2.0s
amount of AGI pled. It's very easy to build the product for the super AGI strong model. The hard thing is figuring out for the current model, how do you elicit the maximum capability? I've never seen anything like the pace you folks at Anthropic are shipping at. We want to remove every single barrier
4.7s
build the product for the super AGI
7.4s
strong model. The hard thing is figuring
9.1s
out for the current model, how do you
11.4s
elicit the maximum capability?
13.6s
I've never seen anything like the pace
15.7s
you folks at Anthropic are shipping at.
17.8s
We want to remove every single barrier
19.6s
to shipping things. The timelines for a lot of our product features have gone down from 6 month to 1 month and sometimes to even one day. You're interviewing hundreds of PMs and you just keep feeling like they're approaching it very incorrectly. The PM role is changing a lot. It's changing really quickly. The thing that
21.8s
lot of our product features have gone
23.8s
down from 6 month to 1 month and
25.7s
sometimes to even one day. You're
27.2s
interviewing hundreds of PMs and you
29.4s
just keep feeling like they're
31.0s
approaching it very incorrectly.
32.2s
The PM role is changing a lot. It's
34.5s
changing really quickly. The thing that
36.4s
is extremely important for building AI native products is iterating so quickly, figuring out a way for you to actually launch features every single week. What do you think are the emerging skills PMs need to develop? It comes back to product taste. As code becomes much cheaper to write, the thing that becomes more valuable is deciding
38.2s
native products is iterating so quickly,
41.0s
figuring out a way for you to actually
42.8s
launch features every single week.
44.5s
What do you think are the emerging
46.1s
skills PMs need to develop?
48.1s
It comes back to product taste. As code
50.7s
becomes much cheaper to write, the thing
52.8s
that becomes more valuable is deciding
54.6s
what to write. Today my guest is Cat Woo, head of product for cloud code and co-work at Enthropic. Cat is at the center of everything that is changing in AI and product and building. And she and her team are building the product that is most changing the way that we all build
56.9s
Today my guest is Cat Woo, head of
58.9s
product for cloud code and co-work at
60.6s
Enthropic. Cat is at the center of
63.0s
everything that is changing in AI and
65.4s
product and building. And she and her
67.4s
team are building the product that is
69.4s
most changing the way that we all build
71.4s
our products. She is so full of insights and wisdom and lessons. This is an episode you cannot miss. Before we get into it, don't forget to check out into it, don't forget to check out lennisprobass.com for an insane set of deals available exclusively to Lenny's newsletter subscribers. With that, I bring you Cat
74.5s
and wisdom and lessons. This is an
77.0s
episode you cannot miss. Before we get
79.3s
into it, don't forget to check out
80.6s
into it, don't forget to check out lennisprobass.com
82.2s
for an insane set of deals available
85.0s
exclusively to Lenny's newsletter
86.8s
subscribers. With that, I bring you Cat
89.3s
subscribers. With that, I bring you Cat Woo.
95.8s
Thanks for having me. I have so many questions. I'm so excited to have you on this podcast. I want to start with giving people an understanding of your role alongside Boris. Uh, everybody knows Boris. This he's His episode is the number one most popular episode on this podcast. No pressure. He uh created Claude Code. He
97.2s
I have so many questions. I'm so excited
99.5s
to have you on this podcast. I want to
101.5s
start with giving people an
103.0s
understanding of your role alongside
105.5s
Boris. Uh, everybody knows Boris. This
108.3s
he's His episode is the number one most
110.2s
popular episode on this podcast. No
111.8s
pressure. He uh created Claude Code. He
114.4s
leads the team, ships uh a bazillion PRs a day from his phone. Just like I don't even know what the number is anymore. I think people don't give you enough credit for the success that Claude Code has had and co-work and all the things you all are building. help us understand
117.1s
a day from his phone. Just like I don't
119.1s
even know what the number is anymore. I
121.2s
think people don't give you enough
123.0s
credit for the success that Claude Code
125.4s
has had and co-work and all the things
126.9s
you all are building. help us understand
129.4s
your role on the team, how you work with Boris, how you split responsibilities, just like what does the PM role look like on on the CloudGo team? I feel very lucky to work with Boris. He's been an amazing thought partner. He's our tech lead. He's very much the product visionary and he is great at
131.4s
Boris, how you split responsibilities,
133.4s
just like what does the PM role look
134.6s
like on on the CloudGo team?
136.2s
I feel very lucky to work with Boris.
137.8s
He's been an amazing thought partner.
139.8s
He's our tech lead. He's very much the
142.2s
product visionary and he is great at
145.6s
setting like this is what the product needs to be in like three months, six months from now. This is like what the AGI pill version of the product is. And a lot of my role is figuring out okay what is the path from where we are today to like that vision 3 to 6 months from
147.3s
needs to be in like three months, six
149.4s
months from now. This is like what the
150.9s
AGI pill version of the product is. And
154.0s
a lot of my role is figuring out okay
156.0s
what is the path from where we are today
157.8s
to like that vision 3 to 6 months from
161.2s
now. And I I spend more of my time on the cross functional. So making sure that our marketing team, sales team, finance, capacity, etc. are like bought in on the plan and that we're all rowing the same direction and that once the feature is ready that there aren't any blockers to shipping it. I think in many
164.4s
the cross functional. So making sure
166.1s
that our marketing team, sales team,
169.4s
finance, capacity, etc. are like bought
171.7s
in on the plan and that we're all rowing
174.6s
the same direction and that once the
177.0s
feature is ready that there aren't any
178.6s
blockers to shipping it. I think in many
180.6s
ways it works well because we kind of like mindmeld but it is actually like remarkably blurry of a line. Like I think we're like 80 mind-l and then there's like this 20 of things that like maybe I care a lot more about them for us. So like I'll drive those and
182.5s
like mindmeld but it is actually like
185.0s
remarkably blurry of a line. Like I
187.3s
think we're like 80 mind-l and then
189.7s
there's like this 20 of things that
191.7s
like maybe I care a lot more about them
193.8s
for us. So like I'll drive those and
195.3s
then like 20 where he cares a lot more than me and he just like drives those. This episode is brought to you by our season's presenting sponsor work OS. What do OpenAI, Anthropic, Cursor, Verscell, Replet, Sierra, Clay, and hundreds of other winning companies all have in common? They are all powered by
197.4s
than me and he just like drives those.
200.1s
This episode is brought to you by our
201.8s
season's presenting sponsor work OS.
204.8s
What do OpenAI, Anthropic, Cursor,
207.4s
Verscell, Replet, Sierra, Clay, and
210.0s
hundreds of other winning companies all
211.7s
have in common? They are all powered by
213.9s
work OS. If you're building a product for the enterprise, you've felt the pain of integrating single signon, skim, arbback, audit, logs, and other features required by large companies. Work OS turns those deal blockers into drop-in APIs with a modern developer platform built specifically for B2B SAS. Literally every startup that I'm an
216.0s
for the enterprise, you've felt the pain
217.8s
of integrating single signon, skim,
220.4s
arbback, audit, logs, and other features
222.8s
required by large companies. Work OS
225.1s
turns those deal blockers into drop-in
227.4s
APIs with a modern developer platform
229.5s
built specifically for B2B SAS.
232.0s
Literally every startup that I'm an
233.9s
investor in that starts to expand up market ends up working with Work OS. And that's because they are the best. Whether you are seedstage startup trying to land your first enterprise customer or a unicorn expanding globally, work OS is the fastest path to becoming enterprise ready and unblocking growth. It's essentially Stripe for enterprise
235.9s
market ends up working with Work OS. And
238.3s
that's because they are the best.
240.0s
Whether you are seedstage startup trying
241.8s
to land your first enterprise customer
243.6s
or a unicorn expanding globally, work OS
246.2s
is the fastest path to becoming
247.8s
enterprise ready and unblocking growth.
250.2s
It's essentially Stripe for enterprise
252.2s
features. Visit works.com to get started or just hit up their Slack where they have actual engineers waiting to answer your questions. Work OS allows you to build faster with delightful APIs, comprehensive docs, and a smooth developer experience. Go to works.com to make your app enterprise ready today. Something that you shared actually
255.0s
or just hit up their Slack where they
256.7s
have actual engineers waiting to answer
258.7s
your questions. Work OS allows you to
260.9s
build faster with delightful APIs,
263.1s
comprehensive docs, and a smooth
264.7s
developer experience. Go to works.com to
267.7s
make your app enterprise ready today.
270.4s
Something that you shared actually
271.8s
before we started recording is the fact that you're interviewing hundreds of PMs all the time. Like if I had a nickel every time someone asked me for an intro to someone at Anthropic to go work at Anthropic as a PM, I'd be I'd be I'd have 30 billion in ARR. It's just like the number one place
274.0s
that you're interviewing hundreds of PMs
276.2s
all the time. Like if I had a nickel
277.8s
every time someone asked me for an intro
279.0s
to someone at Anthropic to go work at
280.5s
Anthropic as a PM, I'd be I'd be I'd
282.6s
have 30 billion in ARR.
285.2s
It's just like the number one place
286.6s
people want to go work at. So, I can only imagine how many PMs you're interviewing. You told me that you're just seeing people doing it, doing it wrong, the way they're approaching what they think it takes to be a successful AIP PM. Talk about what you're seeing and what people need to understand about
288.2s
only imagine how many PMs you're
289.6s
interviewing. You told me that you're
291.8s
just seeing people doing it, doing it
293.8s
wrong, the way they're approaching what
296.0s
they think it takes to be a successful
297.6s
AIP PM. Talk about what you're seeing
299.7s
and what people need to understand about
301.3s
what it is, what it takes to be successful these days. I think before AI, technology shifts were a lot slower. So, you could plan on the 6 to 12 month time horizons. And because you were shipping features at a bit of a slower rate, there was a lot more emphasis on coordinating with all
302.9s
successful these days.
303.9s
I think before AI, technology shifts
306.6s
were a lot slower. So, you could plan on
308.6s
the 6 to 12 month time horizons. And
312.2s
because you were shipping features at a
314.0s
bit of a slower rate, there was a lot
316.2s
more emphasis on coordinating with all
318.6s
the other partner teams to make sure that they're shipping features that unblock your features because code at that time was very expensive to make. Um, I think now with AI and with how much that has accelerated engineering and with how quickly the model capabilities are improving, the timelines for a lot of our product
320.2s
that they're shipping features that
322.3s
unblock your features because code at
324.4s
that time was very expensive to make.
327.0s
Um, I think now with AI and with how
329.8s
much that has accelerated engineering
332.0s
and with how quickly the model
334.0s
capabilities are improving, the
335.8s
timelines for a lot of our product
338.2s
features have gone down from 6 months to one month and sometimes to one week or even one day. And with that, we actually need to make sure that products ship quite quickly. And what that means is as a PM, there should be less emphasis on making sure that you're aligning your like multi-quarter road maps with your
340.8s
one month and sometimes to one week or
342.8s
even one day. And with that, we actually
345.6s
need to make sure that products ship
348.5s
quite quickly. And what that means is as
351.8s
a PM, there should be less emphasis on
355.7s
making sure that you're aligning your
358.2s
like multi-quarter road maps with your
359.9s
partner teams and more emphasis on okay, how can we figure out the fastest way to get something out the door? How can we figure out how to make like a concept corner of our product suite where we can just an engineer has an idea or a PM has an idea and like by the end of the week
362.6s
how can we figure out the fastest way to
365.0s
get something out the door? How can we
367.0s
figure out how to make like a concept
369.7s
corner of our product suite where we can
371.8s
just an engineer has an idea or a PM has
374.9s
an idea and like by the end of the week
377.1s
we are able to get into our users hands. I I think the PMs who do the best on AI native products are are the ones who can figure out how can I like shorten the time from having this idea to actually getting the product in the hands of users and help define what are the most
379.1s
I I think the PMs who do the best on AI
382.5s
native products are are the ones who can
385.4s
figure out how can I like shorten the
387.7s
time from having this idea to actually
389.8s
getting the product in the hands of
391.0s
users and help define what are the most
393.8s
important tasks that need to work out of the box for my product. So, what I love about this is what you're saying is just like people haven't grasped how fast they need to move and what how much of the job now is just moving is helping the team move fast. What what helps do
396.3s
the box for my product. So, what I love
398.2s
about this is what you're saying is just
399.4s
like people haven't grasped how fast
402.6s
they need to move and what how much of
405.0s
the job now is just moving is helping
407.0s
the team move fast. What what helps do
410.6s
that? What do you what do you do? What does your PM team do to help them move this fast other than have access to the the most advanced models? I think the first thing is to set clear queer goals because LMS are so general that actually creates a lot of ambiguity in who we're
412.3s
does your PM team do to help them move
414.0s
this fast other than have access to the
416.2s
the most advanced models? I think the
418.5s
first thing is to set clear queer goals
421.0s
because LMS are so general that actually
423.4s
creates a lot of ambiguity in who we're
425.6s
building for, what problems we're trying to solve, what the top use cases are. And so I think a great PM is able to say, okay, our our key user is professional developers. Uh the main problem that we want to solve for this feature is maybe there's like too many permission prompts and people are
427.4s
to solve, what the top use cases are.
429.1s
And so I think a great PM is able to
431.1s
say, okay, our our key user is
434.2s
professional developers. Uh the main
436.6s
problem that we want to solve for this
438.3s
feature is maybe there's like too many
440.9s
permission prompts and people are
442.1s
feeling fatigue. And like the the use case is we we want professional developers at enterprises to safely get to zero permission prompts. And that actually sets a pretty clear goal because it it rules out a lot of potential approaches for reducing permission prompts so that people can uh get a lot more done with one prompt. And
444.9s
case is we we want professional
448.0s
developers at enterprises to safely get
450.5s
to zero permission prompts. And that
452.2s
actually sets a pretty clear goal
453.6s
because it it rules out a lot of
455.7s
potential approaches for reducing
457.7s
permission prompts so that people can uh
460.2s
get a lot more done with one prompt. And
462.9s
then I think the second thing that's very important is figuring out some repeatable process for getting these features shipped. So uh for cloud code what we do is we actually ship almost all of our features in research preview. We clearly brand this um when we ship something so that users know that this
464.6s
very important is figuring out some
467.3s
repeatable process for getting these
469.5s
features shipped. So uh for cloud code
472.3s
what we do is we actually ship almost
474.2s
all of our features in research preview.
476.3s
We clearly brand this um when we ship
479.1s
something so that users know that this
481.2s
is an early product. This is just an idea. This is just something that we're trying to get feedback on and iterating on and that this might not be supported forever. And what this does is it reduces it reduces our commitment for shipping something. We can just get something out in a week or two. And then
483.2s
idea. This is just something that we're
485.0s
trying to get feedback on and iterating
486.6s
on and that this might not be supported
488.2s
forever. And what this does is it
490.6s
reduces it reduces our commitment for
493.0s
shipping something. We can just get
494.6s
something out in a week or two. And then
497.4s
the third thing that a PM should do is help create the framework for the team so that they know when to pull in cross functional partners and what those crossunctional partners expectations are. So for example, we have a really tight process between engineering, marketing and docs. So when engineers have a feature that they feel is ready
499.8s
help create the framework for the team
501.3s
so that they know when to pull in cross
503.6s
functional partners and what those
505.1s
crossunctional partners expectations
506.6s
are. So for example, we have a really
509.1s
tight process between engineering,
511.6s
marketing and docs. So when engineers
514.3s
have a feature that they feel is ready
516.7s
and that we've dog fooded internally, they post it in our evergreen launch room. And then Sarah who leads our docs and Alex who leads PMM and Tar and Lydia on Devril just like jump in and can turn around the the marketing announcement for it the very next day. And because we have this really tight process it lowers
518.6s
they post it in our evergreen launch
520.2s
room. And then Sarah who leads our docs
523.1s
and Alex who leads PMM and Tar and Lydia
526.3s
on Devril just like jump in and can turn
528.8s
around the the marketing announcement
531.1s
for it the very next day. And because we
533.3s
have this really tight process it lowers
534.8s
the friction for any engineer to ship something and PM is the role that should be setting this up. How do PRDs fit into this? The fact that you said that goals are a really important part just like being aligned on what does success look like? Who is this for? Who's this not for? Are you
536.4s
something and PM is the role that should
538.6s
be setting this up.
539.8s
How do PRDs fit into this? The fact that
541.6s
you said that goals are a really
543.2s
important part just like being aligned
544.4s
on what does success look like? Who is
545.8s
this for? Who's this not for? Are you
547.5s
writing PRDs? Is it just like a couple bullet points? How does how's that evolved in the the world of a BM? So there's two two things that we do. One is we have very rigorous metrics and we do metrics readouts with the entire team every week. The goal of this is to make sure that everyone deeply
549.0s
bullet points? How does how's that
550.3s
evolved in the the world of a BM?
552.1s
So there's two two things that we do.
554.2s
One is we have very rigorous metrics and
557.5s
we do metrics readouts with the entire
559.6s
team every week. The goal of this is to
561.8s
make sure that everyone deeply
563.1s
understands all the facets of our business. What our key goals are, how they're trending, and what drives them. The second thing that we do is we have this list of team principles. And this includes who our key users are, why those are our key users. And the reason that we articulate all of this is so
565.0s
business. What our key goals are, how
567.0s
they're trending, and what drives them.
569.1s
The second thing that we do is we have
571.1s
this list of team principles. And this
574.0s
includes who our key users are, why
576.7s
those are our key users. And the reason
579.5s
that we articulate all of this is so
581.4s
that everybody on the team feels like they understand how our business works. They understand what's important to us and what we're willing to trade off. And it lets people make decisions by themselves without feeling like they're blocked on PM or any other stakeholder. I love how so much of this is like,
583.6s
they understand how our business works.
585.8s
They understand what's important to us
587.4s
and what we're willing to trade off. And
589.5s
it lets people make decisions by
591.0s
themselves without feeling like they're
592.8s
blocked on PM or any other stakeholder.
595.3s
I love how so much of this is like,
596.7s
okay, we still need PMs in the future. There's so much talk of like why do we need PMs? We're just going to ship and build. We need engineers. Oh, we actually do PRD sometimes. So I I think for features that are like particularly ambiguous, it it does help to write out just a one-pager on what
598.6s
There's so much talk of like why do we
599.9s
need PMs? We're just going to ship and
601.4s
build. We need engineers.
603.0s
Oh, we actually do PRD sometimes. So I I
605.8s
think for features that are like
607.0s
particularly ambiguous, it it does help
609.8s
to write out just a one-pager on what
611.9s
the goals are, uh what the delightful use cases are, what the failure modes currently are that we need to fix. And there are occasionally some projects, especially things that require heavy infrastructure that do take many months. And for those situations, we do write PRD still. I want to drill a little bit further
614.8s
use cases are, what the failure modes
617.2s
currently are that we need to fix. And
619.6s
there are occasionally some projects,
621.4s
especially things that require heavy
623.4s
infrastructure that do take many months.
626.2s
And for those situations, we do write
628.0s
PRD still.
629.3s
I want to drill a little bit further
631.0s
into just how you're able to move so fast. I've never seen anything like the pace folks at Anthropic are shipping at like someone made this calendar of launches across Anthropic and it was literally every day there was like a major feature or product. So, one question people had online is uh you
634.0s
fast. I've never seen anything like the
636.5s
pace folks at Anthropic are shipping at
639.1s
like someone made this calendar of
641.6s
launches across Anthropic and it was
643.9s
literally every day there was like a
645.9s
major feature or product. So, one
648.6s
question people had online is uh you
650.8s
guys just launched this uh inc not launch but built this incredible model mythos that is still in preview because it's so powerful people are a little afraid of what it can do. Have you guys been using this? Is this part of the reason you've been able to move so fast? We've been moving pretty fast for
652.8s
launch but built this incredible model
654.6s
mythos that is still in preview because
656.9s
it's so powerful people are a little
658.2s
afraid of what it can do. Have you guys
660.4s
been using this? Is this part of the
661.8s
reason you've been able to move so fast?
663.7s
We've been moving pretty fast for
666.4s
several quarters now. So, I think it it's not fully mythos. Um mythos is an incredibly powerful model. But we do use the models internally and I think this has increased our rate of shipping a little bit but I don't think it explains the bulk bulk of the increase. I I think
669.0s
it's not fully mythos. Um mythos is an
672.1s
incredibly powerful model. But we do use
673.9s
the models internally and I think this
676.6s
has increased our rate of shipping a
678.6s
little bit but I don't think it explains
680.2s
the bulk bulk of the increase. I I think
682.6s
a lot of it is the process and the expectation on the team. So we're very low on process. We want to remove every single barrier to shipping things. We want to make sure every single person on the team feels empowered to take their idea from just an idea to like out in
684.6s
expectation on the team. So we're very
687.0s
low on process. We want to remove every
689.3s
single barrier to shipping things. We
691.3s
want to make sure every single person on
692.8s
the team feels empowered to take their
695.8s
idea from just an idea to like out in
698.1s
the world in less than a week, sometimes even in a day. Cool. Oh man, what a what an advantage to have the best model and also be building product. That's so cool. We are very lucky to be able to work with the Frontier models. Oh my god, what a what an awesome
700.4s
even in a day.
701.4s
Cool. Oh man, what a what an advantage
703.6s
to have the best model and also be
705.4s
building product. That's so cool.
706.7s
We are very lucky to be able to work
708.2s
with the Frontier models.
709.6s
Oh my god, what a what an awesome
711.8s
advantage. Just like build a thing and then use it and then accelerate faster. It's so interesting. There's a couple like these other side things I want to just kind of go on these like side quests on this conversation. There's so much happening with Anthropic and I just I'm so curious to get your insight. One
713.0s
then use it and then accelerate faster.
714.5s
It's so interesting. There's a couple
716.1s
like these other side things I want to
717.4s
just kind of go on these like side
718.6s
quests on this conversation. There's so
720.1s
much happening with Anthropic and I just
721.5s
I'm so curious to get your insight. One
723.9s
is uh a week ago or so the whole source code of cloud code leaked. Somebody got it out there. I think it was a mistake someone made. Is there anything you comment there just like what happened? What went wrong? What should people What went wrong? What should people know? So we immediately looked into this when
727.1s
code of cloud code leaked. Somebody got
729.0s
it out there. I think it was a mistake
730.4s
someone made. Is there anything you
731.8s
comment there just like what happened?
733.0s
What went wrong? What should people
734.6s
What went wrong? What should people know?
735.3s
So we immediately looked into this when
737.6s
we saw it. Um we realized that this was the result of human error. There was um a human working with claw to write uh PR. This was just an update to how we release our packages and it actually went through two layers of human review. And so th this was a result of human
740.7s
the result of human error. There was um
743.2s
a human working with claw to write uh
745.1s
PR. This was just an update to how we
747.7s
release our packages and it actually
751.0s
went through two layers of human review.
753.1s
And so th this was a result of human
755.7s
error and we've hardened our processes to make sure that it doesn't happen in the future. Is this person still at anthropic? Are they doing it right? Yes. Yes. It's it's a process failure and the most important thing is to just like learn from it and to add more safeguards so that doesn't happen again.
757.8s
to make sure that it doesn't happen in
759.4s
the future. Is this person still at
760.9s
anthropic? Are they doing it right?
762.2s
Yes. Yes. It's it's a process failure
764.6s
and the most important thing is to just
767.0s
like learn from it and to add more
768.9s
safeguards so that doesn't happen again.
770.5s
And so that's that's what we've been focused on and most of those have focused on and most of those have shipped. Okay. Uh another question I had is open claw. Uh so recently there's been this move to keep people from using claude subscription with their open clause. People get got really upset. that
772.0s
focused on and most of those have
773.8s
focused on and most of those have shipped.
774.4s
Okay. Uh another question I had is open
777.0s
claw. Uh so recently there's been this
780.9s
move to keep people from using claude
784.1s
subscription with their open clause.
786.7s
People get got really upset. that
788.3s
they're confused why this is happening. It feels like you're there's like, you know, harm caused to the open source community. What what do people what do people need to understand about kind of what went into this decision? So, we've been seeing a lot of demand for quad and we've been working very hard to both scale our infrastructure
789.7s
It feels like you're there's like, you
791.8s
know, harm caused to the open source
793.4s
community. What what do people what do
795.8s
people need to understand about kind of
797.0s
what went into this decision?
798.6s
So, we've been seeing a lot of demand
800.5s
for quad and we've been working very
803.0s
hard to both scale our infrastructure
805.2s
and also to make our harness more token efficient so that you can get more usage out of it. It wasn't designed for third party products which have different uh usage patterns than our first party ones. We spent a bunch of time trying to figure out what is the most seamless transition that we can offer.
807.4s
efficient so that you can get more usage
808.9s
out of it. It wasn't designed for third
811.9s
party products which have different uh
814.6s
usage patterns than our first party
816.0s
ones. We spent a bunch of time
819.8s
trying to figure out what is the most
822.2s
seamless transition that we can offer.
824.3s
And so I was very happy to be able to say that everyone gets some credits alongside their subscription. But yeah, we we did have to make the hard decision that we needed to prioritize our first party products and our API. And so this is this is a decision that resulted from
827.5s
say that everyone gets some credits
829.0s
alongside their subscription. But yeah,
831.2s
we we did have to make the hard decision
832.7s
that we needed to prioritize our first
835.8s
party products and our API. And so this
838.2s
is this is a decision that resulted from
840.7s
that. Yeah, this like to me it makes so much sense. Like you guys are subsidizing this usage at like 200 bucks a month and there's like it's like basically unlimited use of this and like I think people don't understand businesses are trying to make money. We're trying to be profitable here. We
842.6s
much sense. Like you guys are
843.9s
subsidizing this usage at like 200 bucks
847.2s
a month and there's like it's like
849.2s
basically unlimited use of this and like
851.2s
I think people don't understand
852.6s
businesses are trying to make money.
854.1s
We're trying to be profitable here. We
855.9s
can't just like give away compute when it's so in demand. So I get it. Coming back to the PM team, what is just like the PM team look like at Enthropic? How many PMs are there? How are they kind of many PMs are there? How are they kind of organized?
857.6s
it's so in demand. So I get it. Coming
860.2s
back to the PM team, what is just like
862.5s
the PM team look like at Enthropic? How
864.2s
many PMs are there? How are they kind of
865.7s
many PMs are there? How are they kind of organized?
866.6s
Yeah, so we have a few PM teams. Um I think we're maybe around 30 or 40 PMs right now. Uh so we have the research PM team uh who Diane leads and this team is responsible for understanding all of the feedback from our customers for our models and then feeding that to the best
869.8s
think we're maybe around 30 or 40 PMs
871.8s
right now. Uh so we have the research PM
874.2s
team uh who Diane leads and this team is
878.3s
responsible for understanding all of the
880.6s
feedback from our customers for our
882.5s
models and then feeding that to the best
884.7s
research team to act on it and they also shepherd the model launch. Um there's the cloud developer platform team that maintains the APIs that cloud code is built on top of and they also release things like managed agents which is a way for you to build your agents and we can host it on your behalf. And then
886.7s
shepherd the model launch. Um there's
889.4s
the cloud developer platform team that
891.4s
maintains the APIs that cloud code is
893.8s
built on top of and they also release
897.1s
things like managed agents which is a
899.7s
way for you to build your agents and we
901.8s
can host it on your behalf. And then
903.8s
there's cloud code that works on both cloud code and the co-work core products. There's enterprise that helps make cloud code and co-work easier to adopt for all of our enterprise customers. And so this is everything from like cost controls, arvback, security controls and just making sure that these enterprises feel very confident and comfortable uh using using
906.1s
cloud code and the co-work core
907.7s
products. There's enterprise that helps
911.0s
make cloud code and co-work easier to
912.8s
adopt for all of our enterprise
914.8s
customers. And so this is everything
916.9s
from like cost controls, arvback,
919.2s
security controls and just making sure
921.9s
that these enterprises feel very
923.4s
confident and comfortable uh using using
925.9s
our tools and then we also have our growth team that is responsible for growing across our entire product suite. So we work very closely with them on cloud code and co-work growth and I know they also work with um our other teams on C CDP growth. So growth of people who use the cloud API. So speaking of
929.0s
growth team that is responsible for
930.8s
growing across our entire product suite.
932.9s
So we work very closely with them on
934.7s
cloud code and co-work growth and I know
937.0s
they also work with um our other teams
939.3s
on C CDP growth. So growth of people who
942.2s
use the cloud API. So speaking of
944.1s
growth, so Amole was just on the podcast. He had this really interesting insight that most people haven't been sharing. There's always this sense that we need fewer PMs in the future. What's the why do we need PMs? Engineers can just ship. Uh his take is that because engineers are moving so fast, PMs and
945.5s
podcast. He had this really interesting
947.2s
insight that most people haven't been
948.8s
sharing. There's always this sense that
951.0s
we need fewer PMs in the future. What's
952.8s
the why do we need PMs? Engineers can
954.4s
just ship. Uh his take is that because
956.6s
engineers are moving so fast, PMs and
959.6s
designers are squeezed. There's less time to stay on top of everything that is happening. Every there's a feature shipping every day. So his take is he needs more PMs because it's hard to keep up. What's your take there? Do you feel like there will be an increase in hiring of PMs? What do you think is going on
961.4s
time to stay on top of everything that
963.0s
is happening. Every there's a feature
964.4s
shipping every day. So his take is he
966.6s
needs more PMs because it's hard to keep
968.6s
up. What's your take there? Do you feel
970.2s
like there will be an increase in hiring
971.8s
of PMs? What do you think is going on
973.8s
with the PM profession long term? I think all of the roles are merging. PMs are doing some engineering work, engineers are doing PM work, designers are PMing and also landing code. You can either hire a lot more engineers who have great product taste or you can uh keep your engineering hiring the same
975.7s
I think all of the roles are merging.
977.8s
PMs are doing some engineering work,
980.1s
engineers are doing PM work, designers
982.6s
are PMing and also landing code. You can
986.0s
either hire a lot more engineers who
988.6s
have great product taste or you can uh
991.5s
keep your engineering hiring the same
993.0s
and hire a lot more PMs to help guide some of their work. Um on our team we're pretty focused on hiring engineers with great product taste. This this way we can reduce the amount of overhead for shipping any product. Like there are many engineers on our team who are fully
996.5s
some of their work. Um on our team we're
1000.1s
pretty focused on hiring engineers with
1002.6s
great product taste. This this way we
1005.2s
can reduce the amount of overhead for
1007.4s
shipping any product. Like there are
1009.8s
many engineers on our team who are fully
1012.5s
able to end to end go from see user feedback on Twitter through to like ship a product at the end of the week with almost no product involvement. And this I think is actually like the most efficient way to ship something. So I I think like engineer and PM are kind of
1015.8s
feedback on Twitter through to like ship
1017.6s
a product at the end of the week with
1019.2s
almost no product involvement. And this
1021.2s
I think is actually like the most
1022.8s
efficient way to ship something. So I I
1026.6s
think like engineer and PM are kind of
1029.7s
overlapping and you will get a lot of benefit from having more of either. I think product taste is still a very rare skill to have and we'll pretty much hire anyone who we feel has demonstrated this anyone who we feel has demonstrated this strongly. And your background was in engineering, And your background was in engineering, right?
1032.6s
benefit from having more of either. I
1034.7s
think product taste is still a very rare
1038.1s
skill to have and we'll pretty much hire
1041.2s
anyone who we feel has demonstrated this
1045.0s
anyone who we feel has demonstrated this strongly.
1045.8s
And your background was in engineering,
1047.3s
And your background was in engineering, right?
1047.7s
Yeah, I was an engineer for many years. I was then a VC very briefly uh before joining anthropic and actually almost all the PMs on our team have either been engineers or ship code uh here on cloud code and so that that's one of the things that I think helps build trust
1050.0s
I was then a VC very briefly uh before
1053.6s
joining anthropic and actually almost
1056.2s
all the PMs on our team have either been
1058.9s
engineers or ship code uh here on cloud
1062.2s
code and so that that's one of the
1064.1s
things that I think helps build trust
1065.8s
with the team and also just enables us to move a lot faster and then actually our designers also have been front-end engineers before wow because that's that's the big question like there's definitely this merging that's happening the ven diagrams you're combining. I think the big question for a lot of people is if you're coming from engineering or
1069.0s
us to move a lot faster and then
1071.0s
actually our designers also have been
1073.3s
front-end engineers before
1074.9s
wow because that's that's the big
1076.6s
question like there's definitely this
1077.8s
merging that's happening the ven
1079.0s
diagrams you're combining. I think the
1080.9s
big question for a lot of people is if
1082.2s
you're coming from engineering or
1083.4s
product or design, which of those core skills is going to be most valuable? I could see it anthropic and on cloud code, engineering is very valuable. I'm curious if other companies, if you have a design background, becoming a PM is more valuable or just a PMP. I still think it comes back to product
1085.6s
skills is going to be most valuable? I
1087.5s
could see it anthropic and on cloud
1088.8s
code, engineering is very valuable. I'm
1090.6s
curious if other companies, if you have
1092.3s
a design background, becoming a PM is
1094.2s
more valuable or just a PMP.
1096.1s
I still think it comes back to product
1098.7s
taste. Like as code becomes much cheaper to write, the thing that becomes more valuable is deciding what to write. Like what is the right UX for this feature? What is the most delightful way that a user can experience it? What like we we get tens of thousands of GitHub issues asking for every single thing under the
1101.3s
to write, the thing that becomes more
1103.5s
valuable is deciding what to write. Like
1105.8s
what is the right UX for this feature?
1108.2s
What is the most delightful way that a
1110.3s
user can experience it? What like we we
1113.8s
get tens of thousands of GitHub issues
1116.0s
asking for every single thing under the
1118.9s
sun and it takes a lot of care and taste to figure out okay which of these is worth building and what is the right way to build it and I think that that skill set can come from any background but I think that's the most important thing. I think the reason why an engineering
1124.9s
to figure out okay which of these is
1126.6s
worth building and what is the right way
1127.9s
to build it and I think that that skill
1130.7s
set can come from any background but I
1132.3s
think that's the most important thing. I
1133.8s
think the reason why an engineering
1135.6s
background is particularly useful at least for the next few months is if you have an engineering background, you have a better sense for how hard something should be. And that's often a factor in what you choose to build. So like if something is very easy to build, then maybe instead of debating it, you just
1138.7s
least for the next few months is if you
1142.1s
have an engineering background, you have
1144.1s
a better sense for how hard something
1145.7s
should be. And that's often a factor in
1147.9s
what you choose to build. So like if
1149.5s
something is very easy to build, then
1151.8s
maybe instead of debating it, you just
1153.3s
spend an hour doing it. But if something is harder to build and you know that upfront that you know that okay uh this will just like cost a lot more for for our team to get this out the door. So it helps a bit with the prioritization. You said uh in the next for the next few
1155.3s
is harder to build and you know that
1157.1s
upfront that you know that okay uh this
1160.9s
will just like cost a lot more for for
1163.1s
our team to get this out the door. So it
1165.5s
helps a bit with the prioritization.
1167.4s
You said uh in the next for the next few
1169.7s
months is that just like because the models will get so good potentially in the next few months. You may not even need to know that as much. I think the valued skill sets does change quite frequently and so it's really hard to predict more than a few months out. So it's less a commentary on what shift I
1171.8s
models will get so good potentially in
1174.4s
the next few months. You may not even
1175.8s
need to know that as much. I think the
1178.4s
valued skill sets does change quite
1180.8s
frequently and so it's really hard to
1183.6s
predict more than a few months out. So
1185.9s
it's less a commentary on what shift I
1189.2s
think will happen and more of a commentary that I think large shifts will happen. So you're not saying that's when mythos comes out and we'll change everything and that we don't need to know anything about engineering. No, I'm just saying that every every few months it seems like there's a like there's a yeah,
1190.6s
commentary that I think large shifts
1192.6s
will happen.
1193.4s
So you're not saying that's when mythos
1195.0s
comes out and we'll change everything
1196.4s
and that we don't need to know anything
1198.1s
about engineering. No, I'm just saying
1200.1s
that every every few months it seems
1202.0s
like there's a
1202.9s
like there's a yeah,
1203.7s
there's a large increase in coding capability which then changes what other roles are valuable. I think the the most important thing is to be able the most important thing is to be able to to have this like first principles thinking where you can figure out how the tech landscape is changing what the
1206.2s
capability which then changes what other
1209.0s
roles are valuable. I think the
1212.1s
the most important thing is to be able
1215.0s
the most important thing is to be able to
1217.0s
to have this like first principles
1219.1s
thinking where you can figure out how
1222.0s
the tech landscape is changing what the
1225.6s
team really needs from you and to like jump in and fix that hole because I think the work is becoming more amorphous which means that a great PM is able to understand what all the gaps are to figure out what the highest priority ones are and then to just like figure
1228.6s
jump in and fix that hole because I
1232.2s
think the work is becoming more
1234.2s
amorphous which means that a great PM is
1238.5s
able to understand what all the gaps are
1241.6s
to figure out what the highest priority
1243.2s
ones are and then to just like figure
1245.1s
out okay how do I learn that skill set or what is like the skill set that I have that I can like apply to this challenge. So I I think the current environment values people who are who are able to wear a lot of hats are able to swap them and are like very low ego
1247.4s
or what is like the skill set that I
1249.5s
have that I can like apply to this
1251.4s
challenge. So I I think the current
1254.6s
environment values people who are who
1258.2s
are able to wear a lot of hats are able
1260.2s
to swap them and are like very low ego
1263.4s
about what work they do to help the team move faster. I love this answer. There's this question I've been asking people in your in your shoes, folks that are kind of at the bleeding edge of what AI is capable of and building with the latest tools, which is just like where will human
1265.5s
move faster.
1266.3s
I love this answer. There's this
1268.2s
question I've been asking people in your
1270.4s
in your shoes, folks that are kind of at
1271.8s
the bleeding edge of what AI is capable
1273.4s
of and building with the latest tools,
1274.7s
which is just like where will human
1276.5s
brains continue to be useful and necessary for a while until we get to super intelligence. What I'm hearing here is essentially picking the things to work on, knowing where the market's going and figuring out where what to prioritize essentially. And then it's knowing if the thing you've built is good and right
1278.4s
necessary for a while until we get to
1281.0s
super intelligence. What I'm hearing
1283.0s
here is essentially
1285.1s
picking the things to work on, knowing
1287.6s
where the market's going and figuring
1289.3s
out where what to prioritize
1290.7s
essentially. And then it's knowing if
1292.6s
the thing you've built is good and right
1294.4s
and getting it out there in some early version at least. Does that sound right? Is there anything else of just like where human brains will continue to be useful for at least the next few months? I think humans still provide a level of common sense that the models don't. And there's like a thousand moving
1296.3s
version at least. Does that sound right?
1298.0s
Is there anything else of just like
1299.0s
where human brains will continue to be
1301.1s
useful for at least the next few months?
1303.8s
I think humans still provide a level of
1306.6s
common sense that the models don't.
1309.6s
And there's like a thousand moving
1312.2s
pieces to any product launch. Some of them are very small, but there's always a lot that could potentially go wrong. I think the model doesn't always have a great sense of who all the stakeholders are, how they relate to each other, what their preferences are, what are the right venues to communicate with them to
1314.0s
them are very small, but there's always
1316.5s
a lot that could potentially go wrong.
1319.4s
I think the model doesn't always have a
1321.4s
great sense of who all the stakeholders
1324.2s
are, how they relate to each other, what
1326.0s
their preferences are, what are the
1327.7s
right venues to communicate with them to
1329.4s
keep them on board. I think a lot of this like more tacic common sense like EQ kind of knowledge is is still very valuable. Of course, we want the models to get better at this and I think they will be, but right now I think there's still gaps. How do you just kind of deal
1331.1s
this like more tacic common sense like
1334.6s
EQ kind of knowledge is is still very
1337.4s
valuable. Of course, we want the models
1339.9s
to get better at this and I think they
1341.3s
will be, but right now I think there's
1344.0s
still gaps. How do you just kind of deal
1346.2s
as a human going through so much constant change just like just being on the inside of the tornado? Maybe it's calm there, but just like how do you how do you stay on top of what's going on? How do you stay sane through all this craziness that we're moving through?
1348.2s
constant change just like just being on
1350.4s
the inside of the tornado? Maybe it's
1351.8s
calm there, but just like how do you how
1354.2s
do you stay on top of what's going on?
1355.7s
How do you stay sane through all this
1357.9s
craziness that we're moving through?
1359.5s
I think our team is full of people who lean into the chaos. So, we try to face every challenge with a smile because there's always so much going on. There's all there's always so many risks and tricky situations that you know if you get too stressed about anything you'll burn out. And so we really look for
1361.0s
lean into the chaos. So, we try to face
1364.4s
every challenge with a smile because
1366.4s
there's always so much going on. There's
1368.1s
all there's always so many risks and
1371.0s
tricky situations that you know if you
1373.5s
get too stressed about anything you'll
1375.0s
burn out. And so we really look for
1377.2s
people who can kind of like look at a challenge be like that's going to be hard but I'm excited to tackle it and I'm going to do the best that I possibly can and I know I won't be perfect but I'll be able to sleep at night knowing that I did my best. That's an interesting answer to
1379.0s
can kind of like look at a challenge be
1381.1s
like that's going to be hard but I'm
1383.8s
excited to tackle it and I'm going to do
1385.4s
the best that I possibly can and I know
1387.4s
I won't be perfect but I'll be able to
1390.4s
sleep at night knowing that I did my
1391.6s
best. That's an interesting answer to
1393.4s
just like what skills will be important in this future because it's I forget who said this, maybe Ben man that this is the most normal this is the world will ever be. Yeah, it definitely gets harder. Like I feel like there are a lot of weeks where maybe Sunday night there's some like P 0
1396.1s
in this future because it's I forget who
1397.9s
said this, maybe Ben man that this is
1399.4s
the most normal this is the world will
1401.5s
ever be.
1403.2s
Yeah, it definitely gets harder. Like I
1404.8s
feel like there are a lot of weeks where
1408.0s
maybe Sunday night there's some like P 0
1410.3s
and then by Monday there's like a P 0 and by Monday afternoon there's a P 0000 and you're like wow, I can't believe I was so worried about that P 0 from was so worried about that P 0 from Sunday. But I think you just have to acknowledge that there's only so much that you can
1412.8s
and by Monday afternoon there's a P 0000
1415.1s
and you're like wow, I can't believe I
1416.4s
was so worried about that P 0 from
1418.0s
was so worried about that P 0 from Sunday.
1421.0s
But I think you just have to acknowledge
1422.3s
that there's only so much that you can
1423.6s
do that you need to sleep well so that you can make good decisions next day and just like brutally prioritize where you spend your time. What's the most important thing to get right? And be okay letting things go. Like there's there's products that we ship that aren't as polished as I wish they were.
1425.8s
you can make good decisions next day and
1428.7s
just like brutally prioritize where you
1430.3s
spend your time. What's the most
1431.5s
important thing to get right? And be
1433.4s
okay letting things go. Like there's
1435.7s
there's products that we ship that
1437.4s
aren't as polished as I wish they were.
1440.6s
But you know, our our top goal is to help empower professional developers. And if a product isn't successful, as long as it's not blocking the core use case, it's okay because we'll hear the feedback and we'll fix in the next release. Launching a feature that is buggy is the kind of thing that would
1444.6s
help empower professional developers.
1446.5s
And if a product isn't successful, as
1449.8s
long as it's not blocking the core use
1451.6s
case, it's okay because we'll hear the
1455.2s
feedback and we'll fix in the next
1456.5s
release. Launching a feature that is
1459.0s
buggy is the kind of thing that would
1460.3s
have kept me up at night. But it is something that I am now able to like live with knowing that okay, we're going to get that quick feedback and we're going to fix it in the next release. What I'm imagining is there's that gift, I think it's maybe from Pirates of the Caribbean, where it's this guy walking
1464.2s
something that I am now able to like
1466.2s
live with knowing that okay, we're going
1468.5s
to get that quick feedback and we're
1469.9s
going to fix it in the next release.
1471.8s
What I'm imagining is there's that gift,
1473.3s
I think it's maybe from Pirates of the
1474.7s
Caribbean, where it's this guy walking
1476.1s
down a pair of stairs on a ship and the whole ship is just being demolished around him and he's so chill, just strolling down the staircases, everything's falling apart. And that's interesting because everyone I've met through from Anthropic is just so chill and just so like optimistic. Yeah, that's I think that's a really
1478.3s
whole ship is just being demolished
1480.2s
around him and he's so chill, just
1482.1s
strolling down the staircases,
1483.6s
everything's falling apart. And that's
1485.1s
interesting because everyone I've met
1486.5s
through from Anthropic is just so chill
1488.9s
and just so like optimistic.
1491.1s
Yeah, that's I think that's a really
1492.8s
interesting insight is just like having this calmness and optimism versus just like, oh my god, everything's crazy and going going nuts. Yeah, I think if you don't have it, you'll get pretty burnt out. I I think we also tend to hire people who have been in the industry for a while and have experienced lots of ups
1494.7s
this calmness and optimism versus just
1497.1s
like, oh my god, everything's crazy and
1499.0s
going going nuts. Yeah, I think if you
1501.4s
don't have it, you'll get pretty burnt
1503.0s
out. I I think we also tend to hire
1505.5s
people who have been in the industry for
1508.2s
a while and have experienced lots of ups
1510.2s
and downs and have a good sense for what gives them energy and how to maintain their energy over time and I think that's helped us a lot. So interesting. Something that I wanted to ask about is so there's these roles blurring. Engineers are becoming PMs, everyone's dogs are cats, everyone's everyone. What what do we lose in that
1514.8s
gives them energy and how to maintain
1517.3s
their energy over time and I think
1518.9s
that's helped us a lot.
1520.3s
So interesting. Something that I wanted
1522.2s
to ask about is so there's these roles
1523.9s
blurring. Engineers are becoming PMs,
1526.0s
everyone's dogs are cats, everyone's
1527.6s
everyone. What what do we lose in that
1529.8s
in that world? Do we lose like career ladders and clear career paths? Do we lose design consistency, code quality? You know, there's probably some downsides. What are some things you find are just like, okay, that's something we're sacrificing for the greater good. We're sacrificing product consistency. Historically, when code was expensive to write, you would carefully plan out
1532.0s
ladders and clear career paths? Do we
1534.4s
lose design consistency, code quality?
1536.6s
You know, there's probably some
1537.4s
downsides. What are some things you find
1539.3s
are just like, okay, that's something
1540.5s
we're sacrificing for the greater good.
1542.4s
We're sacrificing product consistency.
1545.6s
Historically, when code was expensive to
1548.2s
write, you would carefully plan out
1550.6s
everything in your product suite, how every product relates to each other, what the use case for every single one is, how they integrate, and you would pretty much have one product for each use case. And now with AI moving so quickly and with so many ideas that we need to test out, we do sometimes have
1552.4s
every product relates to each other,
1554.6s
what the use case for every single one
1556.6s
is, how they integrate, and you would
1559.3s
pretty much have one product for each
1561.2s
use case. And now with AI moving so
1564.6s
quickly and with so many ideas that we
1568.1s
need to test out, we do sometimes have
1571.0s
features that overlap with each other. A lot of the times it's because there's two form factors that we love internally and we want to we want the external audience to tell us which one is better. What that means for someone who's a new user though is a new user might not know
1574.2s
lot of the times it's because there's
1576.0s
two form factors that we love internally
1578.2s
and we want to we want the external
1579.9s
audience to tell us which one is better.
1582.7s
What that means for someone who's a new
1585.5s
user though is a new user might not know
1589.2s
okay what is the best path to accomplish X. There is more education we need to do to help people understand what the core features are and what the best practices are for using them. I I think this is the this is the cost of launching a lot of features. Um I think users also feel
1592.2s
X. There is more education we need to do
1596.0s
to help people understand what the core
1598.4s
features are and what the best practices
1600.2s
are for using them. I I think this is
1602.7s
the this is the cost of launching a lot
1606.0s
of features. Um I think users also feel
1609.4s
like it's hard to keep up with the like it's hard to keep up with the latest. Usually in traditional PM you ship a feature every like month or quarter. And so it's really easy for a user to to understand okay I just need to check in on this once a month and I'll learn some
1611.4s
like it's hard to keep up with the latest.
1613.1s
Usually in traditional PM you ship a
1616.2s
feature every like month or quarter. And
1618.5s
so it's really easy for a user to to
1621.5s
understand okay I just need to check in
1623.3s
on this once a month and I'll learn some
1625.0s
new things and if I ignore it for six months it's fine. I don't feel like I'm missing out. I think with these agentic tools, not just called code and co-work, but like across the whole ecosystem, people feel this need to like check Twitter every single day to see what the absolute latest thing is.
1627.0s
months it's fine. I don't feel like I'm
1629.6s
missing out. I think with these agentic
1632.4s
tools, not just called code and co-work,
1634.7s
but like across the whole ecosystem,
1636.9s
people feel this need to like check
1639.3s
Twitter every single day to see what the
1641.6s
absolute latest thing is.
1644.2s
And I think there's more we can do to help people feel less like they're on this ever increasingly fast treadmill and that they feel like I I would love people to feel like they can just open these tools. The tools will educate them um or like teach them what they want to
1647.4s
help people feel less like they're on
1650.2s
this ever increasingly fast treadmill
1654.6s
and that they feel like I I would love
1657.4s
people to feel like they can just open
1658.7s
these tools. The tools will educate them
1661.5s
um or like teach them what they want to
1663.0s
know and that they can just feel more bought along. Yeah, I saw you launch this really interesting feature the other day. I think it's powerup where it basically walks you through all the cool ways and all basically all the best practices to use cloud code. Is that kind of along these lines?
1667.2s
bought along.
1668.1s
Yeah, I saw you launch this really
1669.4s
interesting feature the other day. I
1670.5s
think it's powerup where it basically
1672.6s
walks you through all the cool ways and
1674.1s
all basically all the best practices to
1675.7s
use cloud code. Is that kind of along
1677.0s
these lines?
1677.8s
Yeah, exactly. So, in the past, we didn't actually want to do something like PowerUp because we felt like the product should be intuitive enough that you can that you don't actually need to go through any tutorial. And over time, we've just realized that there's just so many features and there's so much demand for a built-in onboarding experience
1679.8s
didn't actually want to do something
1681.2s
like PowerUp because we felt like the
1682.9s
product should be intuitive enough that
1685.1s
you can that you don't actually need to
1688.0s
go through any tutorial. And over time,
1691.6s
we've just realized that there's just so
1693.3s
many features and there's so much demand
1695.4s
for a built-in onboarding experience
1698.0s
that we we diverged a bit from our original principle saying no no onboarding flow and added this because there's just so many users who wanted to know there's 100 features. What are the 10 that I absolutely need to use? And so we put that together. Yeah, it's such a bizarre world. So Anthropic has been really successful
1700.6s
original principle saying no no
1702.2s
onboarding flow and added this because
1705.3s
there's just so many users who wanted to
1707.3s
know there's 100 features. What are the
1709.6s
10 that I absolutely need to use? And so
1711.5s
we put that together.
1712.5s
Yeah, it's such a bizarre world. So
1714.1s
Anthropic has been really successful
1715.4s
with B2B enterprises where traditionally you don't launch a bunch of stuff. you just kind of have a quarterly release maybe and it's like the opposite of every day we got something new. So just maybe following that thread the run anthropic has been on is just otherworldly. Anthropic was way behind when it started. It was all shared this
1717.8s
you don't launch a bunch of stuff. you
1719.0s
just kind of have a quarterly release
1720.4s
maybe and it's like the opposite of
1722.0s
every day we got something new. So just
1723.8s
maybe following that thread the run
1725.8s
anthropic has been on is just
1728.2s
otherworldly. Anthropic was way behind
1731.1s
when it started. It was all shared this
1733.4s
just like one of the least funded companies. Didn't have distribution. Wasn't the first to go. Openai was way ahead. It was just like no way Anthropic has any chance to compete significantly long term. Now it's just killing it. just beating the biggest companies teams with so much just like the growth is
1735.0s
companies. Didn't have distribution.
1736.6s
Wasn't the first to go. Openai was way
1738.6s
ahead. It was just like no way Anthropic
1740.9s
has any chance to compete significantly
1743.2s
long term. Now it's just killing it.
1745.8s
just beating the biggest companies teams
1748.6s
with so much just like the growth is
1751.2s
just uh like 11 billion dollars in ARR in one month growth by the time this comes out it probably be even higher just being on the inside what what are some ingredients that have allowed Anthropic to be this successful and kind of come from behind and do this well the two most important things are one
1754.4s
in one month growth by the time this
1757.6s
comes out it probably be even higher
1760.5s
just being on the inside what what are
1762.5s
some ingredients that have allowed
1764.2s
Anthropic to be this successful and kind
1766.5s
of come from behind and do this well
1769.4s
the two most important things are one
1772.1s
this unifying mission it's hard to state how important this is. We hire people who care most about bringing safe AGI to all of humanity. And this is actually something that we reference frequently in our decisions about what our entire product or should focus on shipping. And because we put this like mission above
1775.5s
how important this is. We hire people
1779.2s
who care most about bringing safe AGI to
1783.6s
all of humanity. And this is actually
1786.3s
something that we reference frequently
1789.2s
in our decisions about what our entire
1792.4s
product or should focus on shipping. And
1795.3s
because we put this like mission above
1797.4s
any individual product line, we're able to make very fast decisions that cut across the entire org and like execute on them in a unified way. So I think this is this is like something that I've never seen at a company of our scale. And so just to make sure that's clear. So essentially having the the number one
1799.7s
to make very fast decisions that cut
1802.2s
across the entire org and like execute
1804.7s
on them in a unified way. So I think
1807.3s
this is this is like something that I've
1809.8s
never seen at a company of our scale.
1812.3s
And so just to make sure that's clear.
1813.8s
So essentially having the the number one
1815.8s
mission is safety alignment, making sure AI is good for the world. And you're saying just having that as a clear mission makes decisions a lot easier to mission makes decisions a lot easier to make. If there's two competing priorities, we'll talk about which one is more important for Anthropic's mission. And
1818.1s
AI is good for the world. And you're
1819.8s
saying just having that as a clear
1822.2s
mission makes decisions a lot easier to
1824.2s
mission makes decisions a lot easier to make.
1824.5s
If there's two competing priorities,
1826.4s
we'll talk about which one is more
1828.0s
important for Anthropic's mission. And
1830.7s
it makes it a lot easier to decide which of the two we prioritize. And then everyone will stand behind the one that we decide. And so sometimes that means that like, hey, we want to ship something on cloud code, but this other thing is more important. And so we depp prioritize shipping this and we just
1832.9s
of the two we prioritize. And then
1835.5s
everyone will stand behind the one that
1838.0s
we decide. And so sometimes that means
1840.6s
that like, hey, we want to ship
1841.7s
something on cloud code, but this other
1843.3s
thing is more important. And so we depp
1844.9s
prioritize shipping this and we just
1846.2s
wait until later. What's really interesting about that is that explains I think versus another company maybe rhymes with bopen bi uh did a lot of different things and what I'm hearing here essentially is like okay we're not going to launch social network we're not going to launch uh a feed of interesting information because it's not aligned to
1847.6s
interesting about that is that explains
1849.0s
I think versus another company maybe
1851.4s
rhymes with bopen bi uh did a lot of
1854.5s
different things and what I'm hearing
1856.6s
here essentially is like okay we're not
1858.1s
going to launch social network we're not
1859.4s
going to launch uh a feed of interesting
1862.3s
information because it's not aligned to
1864.1s
this mission and and that has kept anthropic focused which is seems to be a core ingredient to the success well when when I think about mission I think about putting anthropics goals ahead of any individual or or any individual product. And so for me, it's I think the second thing that we're very
1866.2s
anthropic focused which is seems to be a
1868.8s
core ingredient to the success
1870.6s
well when when I think about mission I
1872.3s
think about putting anthropics goals
1875.4s
ahead of any individual or or any
1877.6s
individual product. And so for me, it's
1882.8s
I think the second thing that we're very
1884.2s
good at is focus. I think mission to me is slightly different. Mission means is slightly different. Mission means that teams are willing to make sacrifices that hurt their own goals and their own KRs in service of anthropics goals and anthropics KRs. And people are very happy to make those trade-offs. So like an extreme example
1887.0s
is slightly different. Mission means
1889.8s
is slightly different. Mission means that
1891.4s
teams are willing to make sacrifices
1893.8s
that hurt their own goals and their own
1896.0s
KRs in service of anthropics goals and
1898.5s
anthropics KRs. And
1901.7s
people are very happy to make those
1903.4s
trade-offs. So like an extreme example
1906.6s
is if cloud code failed but enthropic succeeded I would be extremely happy and like we're like the whole team is very willing to make decisions that follow that chain of thought. I don't know if you can talk about this in depth but do you feel like the open claw decision is a part of this just
1909.7s
succeeded I would be extremely happy and
1912.4s
like we're like the whole team is very
1914.6s
willing to make decisions that follow
1917.5s
that chain of thought.
1918.8s
I don't know if you can talk about this
1920.6s
in depth but do you feel like the open
1922.4s
claw decision is a part of this just
1923.8s
like okay this is not furthering the mission of enthropic we need to stop this because it's not working in the way we want it to work. I think one of the most important things for Anthropic is to grow the number of users that we're able to reach. One of the ways that
1925.3s
mission of enthropic we need to stop
1928.2s
this because it's not working in the way
1929.9s
we want it to work. I think one of the
1931.7s
most important things for Anthropic is
1933.7s
to grow the number of users that we're
1936.8s
able to reach. One of the ways that
1938.5s
we're able to do this is with the cloud subscriptions with our first party products and so we just very much want to double down on that, but that does come at the expense of third party products sometimes. So we've been talking about cloud, co-work, all these things. Something that I want to make sure people get and
1940.0s
subscriptions with our first party
1941.3s
products and so we just very much want
1944.0s
to double down on that, but that does
1946.2s
come at the expense of third party
1948.0s
products sometimes.
1949.0s
So we've been talking about cloud,
1950.2s
co-work, all these things. Something
1951.4s
that I want to make sure people get and
1953.7s
I'm curious just how you use these tools. So there's cloud code, there's cloud desktop, there's co-work. What's the best way to understand when to use which? When do you use each of these which? When do you use each of these three? So, I tend to use uh cloud code in the terminal when I'm just kicking off like
1954.9s
tools. So there's cloud code, there's
1956.6s
cloud desktop, there's co-work. What's
1960.2s
the best way to understand when to use
1962.3s
which? When do you use each of these
1964.1s
which? When do you use each of these three?
1964.8s
So, I tend to use uh cloud code in the
1967.4s
terminal when I'm just kicking off like
1970.2s
a one-off coding task and I want all of the latest features. Uh the CLI is our initial product surface and it's also the one where our features often land first and so it's the it's the most powerful of all the tools. So that's that's what I tend to use when I'm just
1973.1s
the latest features. Uh the CLI is our
1976.6s
initial product surface and it's also
1978.5s
the one where our features often land
1980.9s
first and so it's the it's the most
1983.8s
powerful of all the tools. So that's
1986.3s
that's what I tend to use when I'm just
1987.9s
like trying to kick off one or like maybe like a handful of tasks at a time. I think desktop really shines when you're doing something that requires front-end work. And so one thing that I love to do is to use our preview feature. So if I'm building a web app, I'll often use Cloud Code and desktop.
1990.2s
maybe like a handful of tasks at a time.
1992.5s
I think desktop really shines when
1994.6s
you're doing something that requires
1996.6s
front-end work. And so one thing that I
1999.0s
love to do is to use our preview
2001.3s
feature. So if I'm building a web app,
2003.5s
I'll often use Cloud Code and desktop.
2006.1s
I'll have the preview pane open on the right hand side so that I can actually see the web app that I'm making in real time as I'm chatting with Claude. It's also really great for people who want something a bit more graphical. Uh, a terminal can feel very unfamiliar to someone who's nontechnical. Um, you get
2007.9s
right hand side so that I can actually
2009.8s
see the web app that I'm making in real
2011.8s
time as I'm chatting with Claude. It's
2013.8s
also really great for people who want
2015.3s
something a bit more graphical. Uh, a
2017.4s
terminal can feel very unfamiliar to
2019.8s
someone who's nontechnical. Um, you get
2022.3s
a bunch of these like scary popups on your machine and you can't click around the way that you're used to in pretty much every other product that you use. So, there's a lot of people who just like don't feel comfortable in terminal. And if that's you, I would highly recommend checking out cloud code on
2024.3s
your machine and you can't click around
2026.8s
the way that you're used to in pretty
2028.2s
much every other product that you use.
2029.8s
So, there's a lot of people who just
2031.7s
like don't feel comfortable in terminal.
2033.7s
And if that's you, I would highly
2035.6s
recommend checking out cloud code on
2037.0s
desktop. Desktop is also great for getting an at a glance view of everything that's happening. So you can see your CLI terminal sessions in desktop. You can see your other desktop sessions. You can see your sessions that you kicked off on web and mobile. So it's a one-stop control plane where you
2039.6s
getting an at a glance view of
2041.7s
everything that's happening. So you can
2043.5s
see your CLI terminal sessions in
2045.7s
desktop. You can see your other desktop
2047.6s
sessions. You can see your sessions that
2049.9s
you kicked off on web and mobile. So
2052.1s
it's a one-stop control plane where you
2054.9s
can see all of your tasks. I think the benefit of web and mobile is that it's really great for kicking things off on the go. So CLI and desktop both require you to be on your local laptop. And this is contravening because sometimes you're out and about, you're like touching grass, you're going on a walk and you
2057.2s
benefit of web and mobile is that it's
2059.8s
really great for kicking things off on
2061.4s
the go. So CLI and desktop both require
2064.5s
you to be on your local laptop. And this
2067.6s
is contravening because sometimes you're
2069.2s
out and about, you're like touching
2070.5s
grass, you're going on a walk and you
2072.4s
don't have your laptop open and you don't I can't I can't count the number of people who I've seen like holding their laptop open like tethered to their phone while they're outside. And this just means that we're missing a product that solves that need. And so for for me, what mobile lets you do is kick off
2073.8s
don't I can't I can't count the number
2076.5s
of people who I've seen like holding
2078.0s
their laptop open like tethered to their
2079.8s
phone while they're outside. And this
2081.8s
just means that we're missing a product
2084.1s
that solves that need. And so for for
2086.7s
me, what mobile lets you do is kick off
2089.2s
these tasks on the go so that you don't you don't need to bring your laptop everywhere and make sure that your laptop's open wherever you are. I love that. I've I've seen people on plane like it's just like such a meme now. Just I need to finish let this agent finish. I can't shut this down. I
2092.0s
you don't need to bring your laptop
2093.3s
everywhere and make sure that your
2095.4s
laptop's open wherever you are.
2097.3s
I love that. I've I've seen people on
2098.8s
plane like it's just like such a meme
2100.3s
now. Just I need to finish let this
2101.7s
agent finish. I can't shut this down. I
2103.3s
need Wi-Fi. And then I think for co-work the the role that this fills is there's a lot of work that everyone does where the output isn't code. So whether that's like getting to Slack zero or inbox zero or whether that's creating a slide deck for some customer meeting that's coming up
2104.2s
And then I think for co-work the the
2106.2s
role that this fills is there's a lot of
2108.4s
work that everyone does where the output
2110.5s
isn't code. So whether that's like
2113.0s
getting to Slack zero or inbox zero or
2115.9s
whether that's creating a slide deck for
2118.8s
some customer meeting that's coming up
2120.5s
or whether that's writing a quick doc on what the goals of a feature are or what the launch plan for a feature is. All these tasks produce outputs that are non-code and co-work is best positioned for that. So the way that I split the products in my mind is if I'm building
2123.3s
what the goals of a feature are or what
2125.9s
the launch plan for a feature is. All
2128.0s
these tasks produce outputs that are
2130.2s
non-code and co-work is best positioned
2132.4s
for that. So the way that I split the
2135.4s
products in my mind is if I'm building
2138.0s
something where the output is code, I'll use cloud code or desktop or cloud code on mobile. And if the output is anything that's not code, I'll use co-work for that's not code, I'll use co-work for it. People are just like sleeping on the success that co-work. It's just like growing incredibly fast and I think
2140.4s
use cloud code or desktop or cloud code
2143.1s
on mobile. And if the output is anything
2145.8s
that's not code, I'll use co-work for
2148.0s
that's not code, I'll use co-work for it.
2148.4s
People are just like sleeping on the
2150.5s
success that co-work. It's just like
2153.2s
growing incredibly fast and I think
2156.0s
people still don't understand maybe what it's for. And so what if you give us a couple use cases just in your work as a PM? What are some like really interesting maybe unexpected ways you use co-work to save you time, get more work done? If you're getting started on co-work, the first thing that you really need to
2158.2s
it's for. And so what if you give us a
2160.9s
couple use cases just in your work as a
2162.6s
PM? What are some like really
2163.8s
interesting maybe unexpected ways you
2165.8s
use co-work to save you time, get more
2167.7s
work done?
2168.6s
If you're getting started on co-work,
2170.6s
the first thing that you really need to
2173.4s
do is connect all the data sources that are relevant to your role because co-work can only do a great job if it has access to all the context that it needs to be able to curate the output for you. So what that means for me is I connect it to my Google calendar. I
2175.8s
are relevant to your role because
2177.4s
co-work can only do a great job if it
2179.3s
has access to all the context that it
2181.4s
needs to be able to curate the output
2183.9s
for you. So what that means for me is I
2186.6s
connect it to my Google calendar. I
2188.2s
connect it to my Slack, to my Gmail, to my Google Drive so that it just knows it has the flexibility to find relevant context to ask questions to pull in threads and this this like substantially improves the quality of the result. The kinds of things I use it for are um like
2190.6s
my Google Drive so that it just knows it
2193.7s
has the flexibility to find relevant
2196.6s
context to ask questions to pull in
2199.2s
threads and this this like substantially
2202.7s
improves the quality of the result. The
2205.2s
kinds of things I use it for are um like
2208.0s
last night I was work where we have this code with cloud conference coming up and there's a few talks that I'm giving there and one of the talks that we're doing talks about the the transition of cloud code from an assistant to like a full-on agent and one of the things that
2210.2s
code with cloud conference coming up and
2212.6s
there's a few talks that I'm giving
2214.0s
there and one of the talks that we're
2216.7s
doing talks about the the transition of
2219.7s
cloud code from an assistant to like a
2222.3s
full-on agent and one of the things that
2225.0s
I wanted to do in this talk was to showcase all of the products that we've been shipping that enable this transition and also to figure out okay what are the what are the success stories that people have had internally that we can use as demos. And so I I have my Google Drive connected, I have
2227.5s
showcase all of the products that we've
2229.6s
been shipping that enable this
2231.5s
transition and also to figure out okay
2234.2s
what are the what are the success
2236.1s
stories that people have had internally
2237.8s
that we can use as demos. And so I I
2241.7s
have my Google Drive connected, I have
2244.3s
Slack connected, um Alex, who's our product marketer, put together like a draft of what the points that we that he thinks we should cover are. And so I just like fed this all into Co-work. I told Co-work the narrative that I want to tell. And it actually just worked for an hour. It it walked through Twitter to
2248.2s
product marketer, put together like a
2250.5s
draft of what the points that we that he
2252.7s
thinks we should cover are. And so I
2254.2s
just like fed this all into Co-work. I
2256.1s
told Co-work the narrative that I want
2257.4s
to tell. And it actually just worked for
2259.4s
an hour. It it walked through Twitter to
2262.6s
see what we launched. It looked through our evergreen launch room. It looked in our Cloud Code announce channel, which is where our team posts demos of what how they've been getting the most value out of Cloud Code. And it synthesized all this together to this 20page deck that I woke up to this morning and I
2264.0s
our evergreen launch room. It looked in
2266.5s
our Cloud Code announce channel, which
2268.3s
is where our team posts demos of what
2271.6s
how they've been getting the most value
2273.3s
out of Cloud Code. And it synthesized
2275.6s
all this together to this 20page deck
2277.5s
that I woke up to this morning and I
2279.5s
read through it and it was like pretty good. There were there were a few tweaks, so I did have to give it a round of feedback. I I like my slides to have extremely minimal words and it was a little too wordy, but you know, it it was far faster than like what I would be
2280.7s
good. There were there were a few
2282.6s
tweaks, so I did have to give it a round
2284.6s
of feedback. I I like my slides to have
2287.1s
extremely minimal words and it was a
2289.5s
little too wordy, but you know, it it
2292.6s
was far faster than like what I would be
2295.1s
able to produce. And because Co-work has access to our whole design system, it actually looks like an anthropic designer put it together. Like it when you visually see it, you're like, Oh, this is like incredibly polished. So, uh these are the kinds of things that are so much faster. like this making
2297.5s
access to our whole design system, it
2299.8s
actually looks like an anthropic
2302.1s
designer put it together. Like it when
2304.5s
you visually see it, you're like, Oh,
2306.7s
this is like incredibly polished. So,
2309.9s
uh these are the kinds of things that
2312.3s
are so much faster. like this making
2314.5s
this slide deck would have taken me hours, but instead it like turns out a draft that is actually quite good so I could focus on making sure that the demos are amazing that we plug into it. This sounds like a dream come true to PMs that putting decks together so PMs that putting decks together so annoying.
2315.9s
hours, but instead it like turns out a
2318.6s
draft that is actually quite good so I
2321.6s
could focus on making sure that the
2322.7s
demos are amazing that we plug into it.
2325.3s
This sounds like a dream come true to
2326.6s
PMs that putting decks together so
2328.6s
PMs that putting decks together so annoying.
2329.4s
It's so slow. I and I love people will see this deck whenever you present this. This will be out in the world to like obviously it's not the the oneshotted version, but you've iterated on it. So just to help people try this for themselves. So step one is connect their what did you say?
2331.4s
I and I love people will see this deck
2333.2s
whenever you present this. This will be
2334.5s
out in the world to like obviously it's
2336.2s
not the the oneshotted version, but
2338.0s
you've iterated on it. So just to help
2340.3s
people try this for themselves. So step
2343.1s
one is connect their what did you say?
2345.5s
Slack. What else do you suggest they Slack. What else do you suggest they connect? Slack, Google calendar, Gmail, G drive. You should connect your communications tools and where you store your source of truth data for what your team cares about, what you care about, and what you're working on. Okay. And then what was the prompt
2347.1s
Slack. What else do you suggest they connect?
2347.8s
Slack, Google calendar, Gmail, G drive.
2351.8s
You should connect your communications
2353.5s
tools and where you store your source of
2356.1s
truth data for what your team cares
2358.6s
about, what you care about, and what
2360.3s
you're working on.
2361.3s
Okay. And then what was the prompt
2363.0s
roughly that you put in there to generate this deck? So I just wrote make me a slide deck for the code with cloud conference. This is what our PMM suggested it should cover. This is the current draft that I made that I don't like. This is one that I made manually that I don't like, but I
2364.8s
generate this deck?
2366.0s
So I just wrote make me a slide deck for
2368.2s
the code with cloud conference. This is
2370.2s
what our PMM suggested it should cover.
2372.8s
This is the current draft that I made
2374.3s
that I don't like. This is one that I
2376.0s
made manually that I don't like, but I
2377.7s
linked it. Can you start by creating a proposed outline with details? Also, make sure it doesn't overlap too much with a keynote talk, which is more important. And then Claude read a bunch of the links that I sent to it and created a proposed outline. So then I read through its proposal and all the
2379.4s
proposed outline with details? Also,
2381.5s
make sure it doesn't overlap too much
2383.0s
with a keynote talk, which is more
2384.5s
important. And then Claude read a bunch
2387.8s
of the links that I sent to it and
2389.9s
created a proposed outline. So then I
2392.9s
read through its proposal and all the
2396.0s
different ideas that it had generated for what we could cover and I just made a decision on what I wanted to actually be in the final deck. And I think this is like an example of what the role of the PM still is today. It's like quad is a great brainstorming partner. It's able
2397.6s
for what we could cover and I just made
2400.1s
a decision on what I wanted to actually
2402.6s
be in the final deck. And I think this
2404.3s
is like an example of what the role of
2405.9s
the PM still is today. It's like quad is
2409.3s
a great brainstorming partner. It's able
2411.8s
to synthesize a massive amount of information really quickly and present all of the possibilities to you. But uh the role of the PM is still to make the end decision of okay what what should belong in the final product. So for this what I ended up deciding was that I wanted the talk to talk to cover the
2414.4s
information really quickly and present
2417.2s
all of the possibilities to you. But uh
2420.0s
the role of the PM is still to make the
2421.7s
end decision of okay what what should
2424.2s
belong in the final product. So for this
2426.2s
what I ended up deciding was that I
2428.9s
wanted the talk to talk to cover the
2431.5s
progression from making local tasks successful to making every PR green to like helping engineers land more PRs and for each of these which demo would be the most compelling and then after this decision about the outline co-work just like went off for a few hours and built the whole slide deck.
2433.6s
successful to making every PR green to
2436.9s
like helping engineers land more PRs and
2439.4s
for each of these which demo would be
2441.0s
the most compelling and then after this
2445.1s
decision about the outline co-work just
2447.1s
like went off for a few hours and built
2449.2s
the whole slide deck.
2450.4s
This is so awesome. What a what an awesome part of the job to not have to do anymore. Uh, and it feels like you're talking to essentially a deck designer that also has like actual knowledge about what you've worked on and and can like make it actually the content what
2452.2s
awesome part of the job to not have to
2454.7s
do anymore. Uh, and it feels like you're
2457.0s
talking to essentially a deck designer
2459.7s
that also has like actual knowledge
2461.5s
about what you've worked on and and can
2464.8s
like make it actually the content what
2467.0s
you want it to be, not just make it look really nice. How did you um how did you do the design system piece? How does that work? How does it know the design system of Anthropic? So what I did for this is we actually already have like a standardized deck that we use across all
2468.6s
really nice. How did you um how did you
2471.4s
do the design system piece? How does
2472.6s
that work? How does it know the design
2474.0s
system of Anthropic? So what I did for
2476.4s
this is we actually already have like a
2479.8s
standardized deck that we use across all
2482.5s
of our external engagements. And so I just gave Claude access to that. And so it's able to see like what colors we use, what fonts we use, the different kinds of what's it called? Like slide formats that are possible. And so it has like 20 of these example slides. Give an example. Got it. So you like
2484.2s
just gave Claude access to that. And so
2486.4s
it's able to see like what colors we
2488.0s
use, what fonts we use, the different
2490.2s
kinds of
2491.8s
what's it called? Like slide formats
2494.2s
that are possible. And so it has like 20
2496.2s
of these example slides.
2497.2s
Give an example. Got it. So you like
2498.6s
upload here's our template work from upload here's our template work from this. Yeah. You can also connect to like your Figma MCP if you if you have your slide format um saved there and it can pull that in. Along those lines, something I'm always curious about is what's kind of in your
2500.3s
upload here's our template work from this.
2500.9s
Yeah. You can also connect to like your
2502.7s
Figma MCP if you if you have your slide
2505.1s
format um saved there and it can pull
2508.1s
that in.
2508.9s
Along those lines, something I'm always
2510.6s
curious about is what's kind of in your
2512.4s
in your stack of tools as a PM and anthropic obviously cloud code and co-work and all the anthropic tools. What else are you using? What are the Slack you mentioned? Is there anything Slack you mentioned? Is there anything else? So my stack is pretty heavily cloud code, co-work. Anthropic largely runs on Slack. Um, I
2514.5s
anthropic obviously cloud code and
2516.8s
co-work and all the anthropic tools.
2518.8s
What else are you using? What are the
2520.3s
Slack you mentioned? Is there anything
2521.7s
Slack you mentioned? Is there anything else?
2522.6s
So my stack is pretty heavily cloud
2525.1s
code, co-work.
2527.6s
Anthropic largely runs on Slack. Um, I
2530.5s
feel like it's like the core OS of our company and day-to-day like a a lot of I I would say maybe 30 of my time is pushing the boundaries of what co-work can do so that I have a very strong sense of what we're not good at. And I spent a lot of time talking with the
2533.0s
company and day-to-day like
2537.0s
a a lot of I I would say maybe 30 of my
2540.3s
time is
2543.2s
pushing the boundaries of what co-work
2546.0s
can do so that I have a very strong
2549.0s
sense of what we're not good at. And
2554.2s
I spent a lot of time talking with the
2555.8s
model to understand why it makes mistakes that it does. We actually have a lot of internal tools that we make. Like I think one of the things that Cloud Code has really unlocked for our entire company is it really lowers the barrier to making any custom app that you want. And so we we've seen this like
2558.5s
mistakes that it does. We actually have
2560.6s
a lot of internal tools that we make.
2563.0s
Like I think one of the things that
2564.3s
Cloud Code has really unlocked for our
2566.5s
entire company is it really lowers the
2569.8s
barrier to making any custom app that
2572.3s
you want. And so we we've seen this like
2575.5s
surge in personalized work software that people are building for like custom use cases instead of um using tools that don't perfectly fit the use case. I got to hear more. What are what are some examples? What are things you've built other people built that are really popular and useful? One of the sales folks on Cloud Code, he
2578.8s
people are building for like custom use
2580.6s
cases instead of um using tools that
2584.6s
don't perfectly fit the use case.
2587.0s
I got to hear more. What are what are
2589.0s
some examples? What are things you've
2590.2s
built other people built that are really
2591.6s
popular and useful?
2593.0s
One of the sales folks on Cloud Code, he
2596.5s
he realized he was making these like repetitive decks over and over and over again. And so he actually has this web app that he built with the examples of the core quad code decks that we know work well. So like a 101, 2011 and mastering quad code. And then he has a way to input specific customer context
2598.1s
repetitive decks over and over and over
2600.1s
again. And so he actually has this web
2603.0s
app that he built with the examples of
2606.0s
the core quad code decks that we know
2608.4s
work well. So like a 101, 2011 and
2610.6s
mastering quad code. And then he has a
2613.1s
way to input specific customer context
2615.5s
that pulls from Salesforce that pulls from gong that pulls from other notes so that we can customize the decks for specific customers. And so it'll pull out things like okay this customer is using like bedrock or cloud called for enterprise or console which affects what features are available to them. Um it
2617.3s
from gong that pulls from other notes so
2619.6s
that we can customize the decks for
2621.4s
specific customers. And so it'll pull
2623.7s
out things like okay this customer is
2626.6s
using like bedrock or cloud called for
2629.3s
enterprise or console which affects what
2631.0s
features are available to them. Um it
2633.2s
will pull out things like okay this customer is concerned about like the code review stage of the SLC. And so we'll add a slide about our code review features there. Um it'll pull out things like okay this customer needs to be like HIPPA compliant or needs XYZ security controls. And so we'll make sure to add
2634.6s
customer is concerned about like the
2637.7s
code review stage of the SLC. And so
2640.6s
we'll add a slide about our code review
2642.7s
features there. Um it'll pull out things
2645.1s
like okay this customer needs to be like
2646.8s
HIPPA compliant or needs XYZ security
2649.6s
controls. And so we'll make sure to add
2651.5s
a slide or two in their deck about that. And then for example, if if this is a customer that's on vertex or bedrock and doesn't want to use cloud for enterprise, then we'll just take out some of the slides that are called for enterprise only features. And so normally this is like manual work that
2654.2s
And then for example, if if this is a
2657.3s
customer that's on vertex or bedrock and
2660.9s
doesn't want to use cloud for
2662.2s
enterprise, then we'll just take out
2663.8s
some of the slides that are called for
2665.5s
enterprise only features. And so
2667.6s
normally this is like manual work that
2669.3s
could take 20 30 minutes or and so people either like spend that time doing it or they'll just decide not to do it and use the general deck. Uh with this it takes like a few seconds and you get a tailored deck. What's interesting about it's like Slack is like the tool that nobody's it's just
2672.1s
or and so people either like spend that
2674.8s
time doing it or they'll just decide not
2676.3s
to do it and use the general deck. Uh
2678.5s
with this it takes like a few seconds
2680.1s
and you get a tailored deck.
2682.3s
What's interesting about it's like Slack
2684.0s
is like the tool that nobody's it's just
2686.6s
like nobody's trying to create their own. Slack just continues to win and it's just like the way you describe it is kind of the OS of so many companies. It's so interesting like people talk about Salesforce as just like SAS. We don't need SAS software anymore. We're going to build our own. It's like Slack
2688.6s
own. Slack just continues to win and
2690.2s
it's just like the way you describe it
2691.6s
is kind of the OS of so many companies.
2693.7s
It's so interesting like people talk
2695.4s
about Salesforce as just like SAS. We
2697.4s
don't need SAS software anymore. We're
2698.5s
going to build our own. It's like Slack
2699.9s
is a durable tool that nobody wants to try to compete with and build a better version. I think it's pretty important communications infrastructure and I think they do the core task of helping everyone get real-time updates incredibly well. Yeah. Like people hate on Slack, but it's really great at what it's trying to
2702.6s
try to compete with and build a better
2704.2s
version. I think it's pretty important
2706.2s
communications infrastructure and I
2708.3s
think they do the core task of helping
2710.5s
everyone get real-time updates
2712.3s
incredibly well.
2713.4s
Yeah. Like people hate on Slack, but
2714.7s
it's really great at what it's trying to
2716.3s
do and like the most cutting edge teams are are hooked on it. So interesting. Yeah. And I also love how custom how easy they've made to customize it. And so it's we we love making Slack bots and th this kind of like hackability uh means that we're able to integrate with
2719.3s
are are hooked on it. So interesting.
2721.3s
Yeah. And I also love how custom how
2723.8s
easy they've made to customize it. And
2726.0s
so it's we we love making Slack bots and
2730.5s
th this kind of like hackability uh
2733.2s
means that we're able to integrate with
2734.6s
Slack the way that we want to. So really appreciate Slack's work on that. Time time to buy some CRM stock. I am so excited to tell you about this season's supporting sponsor, Vanta. Vanta helps over 15,000 companies like Cursor, Ramp, Dualingo, Snowflake, and Atlassian earn and prove trust with their customers. Teams are building and shipping products
2736.4s
appreciate Slack's work on that.
2737.9s
Time time to buy some CRM stock. I am so
2741.2s
excited to tell you about this season's
2743.0s
supporting sponsor, Vanta. Vanta helps
2745.6s
over 15,000 companies like Cursor, Ramp,
2749.4s
Dualingo, Snowflake, and Atlassian earn
2752.5s
and prove trust with their customers.
2755.0s
Teams are building and shipping products
2756.7s
faster than ever thanks to AI. But as a result, the amount of risk being introduced into your product and your business is higher than it's ever been. Every security leader that I talk to is feeling the increasing weight of protecting their organization, their business, and not to mention their customer data. Because things are moving
2759.0s
result, the amount of risk being
2760.7s
introduced into your product and your
2762.6s
business is higher than it's ever been.
2765.0s
Every security leader that I talk to is
2767.0s
feeling the increasing weight of
2769.4s
protecting their organization, their
2771.3s
business, and not to mention their
2772.9s
customer data. Because things are moving
2775.0s
so fast, they are constantly reacting, having to guess at priorities, and having to make do with outdated solutions. Vanta automates compliance and risk management with over 35 security and privacy frameworks including SOCK 2, ISO 27,0001 and HIPPA. This helps companies get compliant fast and stay compliant more than ever before. Trust has the power to make or
2777.6s
having to guess at priorities, and
2779.3s
having to make do with outdated
2780.9s
solutions. Vanta automates compliance
2783.1s
and risk management with over 35
2785.6s
security and privacy frameworks
2787.6s
including SOCK 2, ISO 27,0001 and HIPPA.
2791.4s
This helps companies get compliant fast
2793.7s
and stay compliant more than ever
2795.8s
before. Trust has the power to make or
2798.0s
break your business. Learn more at break your business. Learn more at vanta.comlenny. And as a listener of this podcast, you get 1,000 off Vanta. That's get 1,000 off Vanta. That's vanta.comlenny. Okay. Uh so you talked about all these different teams that and how they use cloud code and co-work to operate. Which teams do you find other than
2800.0s
break your business. Learn more at vanta.comlenny.
2802.1s
And as a listener of this podcast, you
2803.8s
get 1,000 off Vanta. That's
2806.4s
get 1,000 off Vanta. That's vanta.comlenny.
2809.0s
Okay. Uh so you talked about all these
2810.5s
different teams that and how they use
2812.6s
cloud code and co-work to operate. Which
2815.1s
teams do you find other than
2816.2s
engineering? I imagine engineering is the biggest token spender, but if not that'd be really interesting. What what's kind of like the second place function right now for tokens? Oh, applied AI is amazing at pushing the boundaries of what quad code and co-work can do. A a lot of our applied AI team spends time with our customers helping
2817.7s
the biggest token spender, but if not
2819.8s
that'd be really interesting. What
2821.0s
what's kind of like the second place
2822.2s
function right now for tokens?
2824.5s
Oh, applied AI is amazing at pushing the
2827.7s
boundaries of what quad code and co-work
2829.8s
can do. A a lot of our applied AI team
2834.2s
spends time with our customers helping
2836.6s
them adopt our API. And so sometimes our applied team will for example make prototypes on behalf of these customers which cloud code makes so much faster than it used to be. They they also have the dual goal of needing to manage a lot of customer coms, a lot of like customer inbound and historical context call
2840.2s
applied team will for example make
2841.6s
prototypes on behalf of these customers
2843.9s
which cloud code makes so much faster
2846.3s
than it used to be. They they also have
2849.1s
the dual goal of needing to manage a lot
2852.6s
of customer coms, a lot of like customer
2855.2s
inbound and historical context call
2858.2s
notes. And so they're both extremely heavy on co-work and on cloud code. And just to understand applied AI, is that like is that like forward to play engineering sort of role? Like what do they how would you how would most people describe what applied the applied AI team is doing? Yeah, it's helping our
2860.8s
heavy on co-work and on cloud code.
2862.7s
And just to understand applied AI, is
2864.2s
that like is that like forward to play
2866.2s
engineering sort of role? Like what do
2867.8s
they how would you how would most people
2869.5s
describe what applied the applied AI
2871.2s
team is doing? Yeah, it's helping our
2873.4s
customers adopt the latest API and uh model features um across their company both for powering their company's products and also for internal products and also for internal acceleration. Got it. So it's like customer success go to markety kind of like for deploy engineering sort of. Exactly. It's like a very technical go to market person.
2877.4s
model features um across their company
2879.9s
both for powering their company's
2882.2s
products and also for internal
2884.6s
products and also for internal acceleration.
2885.4s
Got it. So it's like customer success go
2887.3s
to markety kind of like for deploy
2889.4s
engineering sort of.
2890.4s
Exactly. It's like a very technical go
2892.1s
to market person.
2893.0s
Got it. Okay. Awesome. So that's so you're saying that might be the second uh org that uses the most tokens. Yeah. And then we we also see them pushing the boundaries of what co-work can do. So for example, if so a lot of these folks cover multiple customers and in any given day can have like five to
2894.7s
you're saying that might be the second
2896.6s
uh org that uses the most tokens.
2899.8s
Yeah. And then we we also see them
2901.8s
pushing the boundaries of what co-work
2903.6s
can do. So for example, if so a lot of
2907.7s
these folks cover multiple customers and
2910.8s
in any given day can have like five to
2913.7s
10 customer engagements on a high day. And so what they often use co-work to do is the night before they'll ask it to summarize, okay, what are all my customer meetings that are coming up the next day? um what are all the what are all the things that this customer has asked me
2916.7s
And so what they often use co-work to do
2919.3s
is the night before they'll ask it to
2921.8s
summarize, okay, what are all my
2923.2s
customer meetings that are coming up the
2924.6s
next day?
2925.8s
um what are all the what are all the
2927.7s
things that this customer has asked me
2929.4s
for uh what's top of mind for them what are the action items from the past meetings and co-work will just put together this like dossier this like brief of what they should be aware of going into the next meeting and co-work can also research answers so if if a customer asked okay when is feature X
2931.8s
are the action items from the past
2933.0s
meetings and co-work will just put
2935.9s
together this like dossier this like
2937.9s
brief of what they should be aware of
2939.8s
going into the next meeting and co-work
2941.8s
can also research answers so if if a
2944.5s
customer asked okay when is feature X
2946.6s
going to launch um co-work can help the pi person research through Slack to get the latest ETA add that to the add that to the notes so that during the customer call the pi person has the absolute latest and these are just workflows that people are building for themselves and sharing with other people on their team.
2949.1s
pi person research through Slack to get
2952.0s
the latest ETA add that to the add that
2955.0s
to the notes so that during the customer
2957.3s
call the pi person has the absolute
2959.6s
latest and these are just workflows that
2961.6s
people are building for themselves and
2963.4s
sharing with other people on their team.
2965.2s
So cool something that kind of this question this trend uh I don't know question topic comes up a lot recently which is um token spend exceeding people's salary where people just use AI and it costs more than how much they're making. Are there any numbers floating around anthropic of just like how much tokens spend say engineers
2967.4s
question this trend uh I don't know
2969.8s
question topic comes up a lot recently
2971.8s
which is um token spend exceeding
2974.6s
people's salary where people just use AI
2978.3s
and it costs more than how much they're
2979.8s
making. Are there any numbers floating
2981.6s
around anthropic of just like how much
2983.5s
tokens spend say engineers
2986.3s
uh spend I don't know a month a day PMs anything like that it is clear to us that as the models get better people delegate far more tasks to it and they spend a lot more hours in tools like quad code and co-work and so we do see the token cost per engineer or
2989.1s
anything like that
2990.2s
it is clear to us that as the models get
2992.5s
better people delegate far more tasks to
2995.5s
it and they spend a lot more hours in
2998.3s
tools like quad code and co-work and so
3001.2s
we do see the token cost per engineer or
3004.9s
like per any knowledge worker increase every time that there's a model jump or like a substantial product improvement. I think it's still much lower than what the average engineer salary is, but we see the percentage increasing over time. It's such an interesting like we talked about how you have access to the most
3007.8s
every time that there's a model jump or
3010.1s
like a substantial product improvement.
3012.6s
I think it's still much lower than what
3016.3s
the average engineer salary is, but we
3019.6s
see the percentage increasing over time.
3021.9s
It's such an interesting like we talked
3023.4s
about how you have access to the most
3025.3s
cutting edge models and other advantage of working anthropic. I I believe you guys have basically unlimited tokens. You don't you can use as much as you want. Is that right? We can use a lot of tokens. Some people do run into limits. So, okay, there's a limit. Okay, Baris, shut it down. H, okay. Like, it's so
3026.7s
of working anthropic. I I believe you
3029.0s
guys have basically unlimited tokens.
3030.7s
You don't you can use as much as you
3032.0s
want. Is that right?
3033.0s
We can use a lot of tokens. Some people
3035.1s
do run into limits. So,
3036.8s
okay, there's a limit. Okay, Baris, shut
3039.2s
it down. H, okay. Like, it's so
3042.2s
interesting how many advantages come from having the most advanced model. It's such an interesting like flywheel that starts to kick in. I think we also believe a lot in empowering our internal teams to build as fast as possible. And we also trust that everyone understands how much capacity that serving these models truly costs. and we trust our
3043.8s
from having the most advanced model.
3045.7s
It's such an interesting like flywheel
3047.8s
that starts to kick in. I think we also
3050.3s
believe a lot in empowering our internal
3053.0s
teams to build as fast as possible. And
3056.5s
we also trust that everyone understands
3059.5s
how much capacity that serving these
3062.7s
models truly costs. and we trust our
3065.4s
team to use the tokens responsibly. So, it's very frowned upon to waste tokens, but we do trust individuals to make that judgment call. Awesome. Coming back to the PM role, you talked we talked a little bit about this, but I think this will be really interesting for people to hear. Just
3067.8s
it's very frowned upon to waste tokens,
3071.1s
but we do trust individuals to make that
3073.3s
judgment call.
3074.2s
Awesome. Coming back to the PM role, you
3077.4s
talked we talked a little bit about
3078.6s
this, but I think this will be really
3079.8s
interesting for people to hear. Just
3082.6s
what I want to understand is what do you think are the kind of the emerging skills that PMs need to develop slash you most look for AI companies most look for when they're hiring PMs these days? I think the hardest skill is being able to define what the product should look like
3084.0s
think are the kind of the emerging
3085.9s
skills that PMs need to develop slash
3090.1s
you most look for AI companies most look
3092.6s
for when they're hiring PMs these days?
3095.2s
I think the hardest skill is
3098.8s
being able to
3101.4s
define what the product should look like
3103.8s
a month from now. I think there's a lot of ambiguity and what models are capable of in that timeline and how user behavior will change. But I think there are patterns that the best PMs can see based on how users are abusing the limits of the existing product and the best PMS can sense that
3106.2s
of ambiguity and what models are capable
3108.6s
of in that timeline and how user
3110.4s
behavior will change.
3112.5s
But I think there are patterns that the
3115.0s
best PMs can see based on how users are
3118.3s
abusing the limits of the existing
3120.2s
product and the best PMS can sense that
3124.2s
can set a direction and can steadily execute towards it and change the path if the model capabilities are much better than or worse than what they had originally expected. I think it is very hard to be the right amount of AGI pilled because I think everyone can see this like this future where the models
3127.0s
execute towards it and change the path
3129.8s
if the model capabilities are much
3131.8s
better than or worse than what they had
3134.0s
originally expected. I think it is very
3136.6s
hard to be the right amount of AGI
3138.6s
pilled because I think everyone can see
3141.0s
this like this future where the models
3144.5s
are extremely smart and can do almost everything in which case you actually don't need that complicated a product. You can actually just have a text box again where you tell the model what you want. And it's so smart that it can add any tool or add any integration that it
3146.9s
everything in which case you actually
3149.8s
don't need that complicated a product.
3151.7s
You can actually just have a text box
3153.1s
again where you tell the model what you
3154.7s
want. And it's so smart that it can add
3158.6s
any tool or add any integration that it
3160.9s
needs to like get the job done. It knows when it's uncertain. and they can ask clarifying questions like it's kind of very easy to build the product for the super AGI uh strong model. I think the hard thing is figuring out for the current model. How do you elicit the maximum capability? How do you help
3163.0s
when it's uncertain. and they can ask
3164.5s
clarifying questions like it's kind of
3167.4s
very easy to build the product for the
3169.7s
super AGI uh strong model. I think the
3173.8s
hard thing is figuring out for the
3176.7s
current model. How do you elicit the
3180.0s
maximum capability? How do you help
3182.4s
maximum capability? How do you help users go get onto the the golden path? How do you like guide users to interact with the model's strengths and like patch its weaknesses? Th this skill is like pretty rare. And how do you build that skill? Is it just using each like basically understanding the limits of each model
3184.1s
go get onto the
3187.3s
the golden path? How do you like guide
3189.9s
users to interact with the model's
3192.3s
strengths and like patch its weaknesses?
3195.4s
Th this skill is like pretty rare.
3199.1s
And how do you build that skill? Is it
3200.5s
just using each like basically
3202.4s
understanding the limits of each model
3204.3s
having like you talked about taste, understanding having taste into what the model maybe is capable of, what it's great and not great at, where it's great and not great at, where it's changed. I think it's spending a ton of time talking and using the model. One of the things I really like to do is to ask the
3206.6s
understanding having taste into what the
3208.8s
model maybe is capable of, what it's
3210.2s
great and not great at, where it's
3211.4s
great and not great at, where it's changed.
3212.0s
I think it's spending a ton of time
3213.8s
talking and using the model. One of the
3216.8s
things I really like to do is to ask the
3219.8s
model to introspect on its own behaviors. So sometimes when I notice that the model does something unexpected, like for example, there's like situations where the model will make a front-end change and run tests but not actually use the UI. It's actually pretty useful to ask the model to reflect on why it did this. And
3221.2s
behaviors. So sometimes when I notice
3224.2s
that the model does something
3225.6s
unexpected, like for example, there's
3228.6s
like situations where the model will
3231.5s
make a front-end change and run tests
3234.2s
but not actually use the UI. It's
3236.6s
actually pretty useful to ask the model
3239.7s
to reflect on why it did this. And
3242.9s
sometimes they'll say that hey there was like something confusing in the system prompt or I didn't realize that um the front-end verification was like part of this task or hey I delegated the verification to this sub agent and the sub agent didn't do the test and I didn't check its work. A lot of times
3244.5s
like something confusing in the system
3245.8s
prompt or I didn't realize that um the
3249.6s
front-end verification was like part of
3251.0s
this task or hey I delegated the
3253.3s
verification to this sub agent and the
3255.0s
sub agent didn't do the test and I
3257.0s
didn't check its work. A lot of times
3259.7s
just like being very curious about why the model made the decision that it did will show you what misled it so that you can fix the harness in order to close this gap. The other thing that helps is to figure out who the taste who are the users who you trust the most to give you accurate
3262.1s
the model made the decision that it did
3264.1s
will show you
3266.2s
what misled it so that you can fix the
3268.6s
harness in order to close this gap. The
3271.8s
other thing that helps is to figure out
3275.0s
who the taste who are the users who you
3278.2s
trust the most to give you accurate
3280.3s
feedback about the model. Usually there's like a handful of people who are much better than others at articulating what makes a specific model or model harness combination good. And there's a lot of people who will give you feedback, but not everyone's feedback is as qualified. And so finding a group of those like five people you trust is
3282.9s
there's like a handful of people who are
3285.1s
much better than others at articulating
3287.5s
what makes a specific model or model
3289.4s
harness combination good. And there's a
3293.1s
lot of people who will give you
3294.6s
feedback, but not everyone's feedback is
3296.2s
as qualified. And so finding a group of
3299.3s
those like five people you trust is
3301.6s
really important for getting very fast feedback. I think the third thing that is useful but not everyone loves doing is building evals. You don't need to build hundreds of evals for them to be useful. Just building 10 great evals is important for helping the team quantify what the goal is and what their progress
3303.6s
feedback. I think the third thing that
3306.7s
is useful but not everyone loves doing
3309.3s
is building evals. You don't need to
3312.2s
build hundreds of evals for them to be
3315.3s
useful. Just building 10 great evals is
3319.2s
important for helping the team quantify
3322.7s
what the goal is and what their progress
3324.8s
towards it is and what they're missing. And so I think eval is this like underappreciated thing that more more PMs more engineers should be working on. We've covered evals a bunch. There's this trend of just like that is the future of product management is writing evals because it and essentially it's what does success look like? Okay, cool.
3327.1s
And so I think eval is this like
3329.0s
underappreciated thing that more more
3331.2s
PMs more engineers should be working on.
3333.7s
We've covered evals a bunch. There's
3335.6s
this trend of just like that is the
3337.0s
future of product management is writing
3338.2s
evals because it and essentially it's
3340.2s
what does success look like? Okay, cool.
3341.7s
Let me actually concretely define it and then we'll know. How much of your time are you spending writing evals would you are you spending writing evals would you say? I I think the importance of evals varies a bit based on the feature that you're working on and or like what the problem
3343.2s
then we'll know. How much of your time
3344.6s
are you spending writing evals would you
3346.4s
are you spending writing evals would you say?
3346.7s
I I think the importance of evals varies
3348.8s
a bit based on the feature that you're
3350.7s
working on and or like what the problem
3354.1s
you're trying to solve is. So there are a lot of folks on our team who do spend a lot of time working on eval. have a small pod of folks who collaborate very closely with research to more precisely understand our quad code behaviors and what the largest areas of improvement are and trying to measure those pretty
3356.4s
a lot of folks on our team who do spend
3358.4s
a lot of time working on eval. have a
3360.6s
small pod of folks who collaborate very
3363.7s
closely with research to more precisely
3366.1s
understand our quad code behaviors and
3370.5s
what the
3372.5s
largest areas of improvement are and
3374.2s
trying to measure those pretty
3375.4s
concretely. I personally jump into evals when there's a feature that I think needs a bit more product definition and often the output of this is okay here are like five evas that I made um this is how you run them these are the ones that succeed and these are the ones that
3378.4s
when there's a feature that I think
3381.2s
needs a bit more product definition and
3384.1s
often the output of this is okay here
3387.6s
are like five evas that I made um this
3390.6s
is how you run them these are the ones
3392.4s
that succeed and these are the ones that
3393.8s
don't and this is like the prompt that I've used to increase the success rate it varies a lot though based on the exact feature uh not every feature needs it but I think features such as memory benefit a lot from this uh point you made about people being very good at evaluating models so interesting. It's
3395.8s
I've used to increase the success rate
3399.0s
it varies a lot though based on the
3401.0s
exact feature uh not every feature needs
3403.1s
it but I think features such as memory
3405.1s
benefit a lot from this uh point you
3407.5s
made about people being very good at
3409.4s
evaluating models so interesting. It's
3410.8s
almost like a human eval of just like okay they understand where it's spiking or it's maybe lacking. Uh is there anyone specific that you want to shout out that's very good at this? Uh two people who I think are incredible at this are um one Amanda who def who molds Claude's character. It's just like
3413.0s
okay they understand where it's spiking
3414.9s
or it's maybe lacking. Uh is there
3417.4s
anyone specific that you want to shout
3419.4s
out that's very good at this?
3420.8s
Uh two people who I think are incredible
3423.2s
at this are um one Amanda who def who
3427.9s
molds Claude's character. It's just like
3431.0s
such a hard role because the task is so ambiguous. Even coding is easier because you can verify the success whereas crafting the character requires a very strong sense of conviction in what who Claude should be. And I think she has like an incredible ability to not only mold the character, but also to like
3435.4s
ambiguous. Even coding is easier because
3438.6s
you can verify the success whereas
3440.9s
crafting the character requires a very
3443.0s
strong sense of conviction in what who
3446.6s
Claude should be. And I think she has
3449.0s
like an incredible ability to not only
3452.0s
mold the character, but also to like
3453.9s
articulate what the goals are, what the character, what's successful and what's not. The other group of people who I really trust is just like the Cloud Code team. Um, so we often have team lunches and whenever there's a new model we're testing. One of the fastest ways for us to get feedback is to just like at these
3457.2s
character, what's successful and what's
3459.9s
not. The other group of people who I
3463.1s
really trust is just like the Cloud Code
3465.0s
team. Um, so we often have team lunches
3467.9s
and whenever there's a new model we're
3469.4s
testing. One of the fastest ways for us
3471.7s
to get feedback is to just like at these
3474.0s
team lunches just like go to every single person and just be like, Hey, what is your vibe on the model? And what is your vibe on the model? And oftentimes we'll we'll get feedback like, Okay, this model is like not fully explaining its thinking. It's like too abrupt. or like hey this model's like um just like
3475.4s
single person and just be like, Hey,
3477.0s
what is your vibe on the model? And
3479.4s
what is your vibe on the model? And oftentimes
3481.0s
we'll we'll get feedback like, Okay,
3482.8s
this model is like not fully explaining
3485.7s
its thinking. It's like too abrupt. or
3488.2s
like hey this model's like um just like
3492.1s
loves writing a ton of memories but like we're not sure if the memories are high quality or not or like some people will notice that okay this this model loves to test itself which is great or like this model isn't testing itself enough. So that informs what data we look at to
3494.2s
we're not sure if the memories are high
3495.8s
quality or not or like some people will
3498.3s
notice that okay this this model loves
3500.7s
to test itself which is great or like
3502.6s
this model isn't testing itself enough.
3504.9s
So that informs what data we look at to
3507.3s
verify okay is this a larger pattern. So we we have a ton of data but it is very hard to extract insights and so the the feedback from this group helps us inform okay what are the hypotheses we want to test and then we're able to extract uh data to uh test that
3510.5s
we we have a ton of data but it is very
3513.2s
hard to extract insights and so the the
3517.0s
feedback from this group helps us inform
3518.7s
okay what are the hypotheses we want to
3520.2s
test and then we're able to extract uh
3523.6s
data to uh test that
3525.8s
this point you made about the character of Claude I had Ben man on the podcast co-founder and he talked about this just like the character the constitution of Claude is such an important part of of of Claude and I I didn't realize until afterwards just Like like people like with open claw actually one of the examp
3527.1s
of Claude I had Ben man on the podcast
3529.6s
co-founder and he talked about this just
3531.8s
like the character the constitution of
3533.4s
Claude is such an important part of of
3535.7s
of Claude and I I didn't realize until
3538.6s
afterwards just Like like people like
3541.0s
with open claw actually one of the examp
3542.6s
one one of the reasons people are sad is like the personality of your claw is like because Claude's personality is so good and fun and and interesting unlike other models and there's and the way he put it is the personality is what makes Claude so good at so many things. It feels like this like trivial side thing.
3543.9s
like the personality of your claw is
3546.6s
like because Claude's personality is so
3549.3s
good and fun and and interesting unlike
3552.3s
other models and there's and the way he
3554.3s
put it is the personality is what makes
3556.4s
Claude so good at so many things. It
3558.6s
feels like this like trivial side thing.
3560.6s
Okay, it's going to be funny and interesting and talk in a fun way but it's like so core to the success of Claude. Is there anything you get there about just like what people may not understand about why the character as you described and the personality is so you described and the personality is so key?
3561.8s
interesting and talk in a fun way but
3564.1s
it's like so core to the success of
3566.4s
Claude. Is there anything you get there
3568.1s
about just like what people may not
3569.4s
understand about why the character as
3571.6s
you described and the personality is so
3573.2s
you described and the personality is so key?
3574.2s
When you reflect on everyone you've worked with, there's just some people where you're like, I really like their energy. Like, I really like their vibe. And when people think about Quad and Quad Code, this is one of the things that people bring up the most where they just really love that COD is like it's
3575.5s
worked with, there's just some people
3577.7s
where you're like, I really like their
3579.8s
energy. Like, I really like their vibe.
3582.4s
And when people think about Quad and
3584.7s
Quad Code, this is one of the things
3587.1s
that people bring up the most where they
3589.7s
just really love that COD is like it's
3593.1s
it's like lighthearted and fun. Um, but it also is extremely competent at your task. People really like that Claude's low ego. And so if you tell it, hey, you did this thing wrong. It's like truly sorry. It's like, oh shoot, like, thanks for telling me. Like, let me fix it. Let's work together. It's also very
3596.8s
it also is extremely competent at your
3599.7s
task. People really like that Claude's
3602.2s
low ego. And so if you tell it, hey, you
3604.4s
did this thing wrong. It's like truly
3605.9s
sorry. It's like, oh shoot, like, thanks
3608.3s
for telling me. Like, let me fix it.
3609.8s
Let's work together. It's also very
3611.7s
positive. So if you're feeling like, oh, this is like an insurmountable task. I don't know h how to get started. Quad is like, okay, it's okay. The these are like the steps that I think we should take. like, do you want me to get started on it for you? I think part of
3614.8s
this is like an insurmountable task. I
3616.8s
don't know h how to get started. Quad is
3619.9s
like, okay, it's okay. The these are
3622.6s
like the steps that I think we should
3624.1s
take. like, do you want me to get
3625.9s
started on it for you? I think part of
3628.8s
what makes a great co-orker is this positivity, this like bias towards action, this this ability to give you like earnest feedback, not just agreeing with every single thing that you say. And so we try to imbue this into cloud because we think it makes it a lot more enjoyable to work with.
3631.5s
positivity, this like bias towards
3633.4s
action, this this ability to give you
3636.6s
like earnest feedback, not just agreeing
3639.3s
with every single thing that you say.
3640.9s
And so we try to imbue this into cloud
3642.8s
because we think it makes it a lot more
3644.3s
enjoyable to work with.
3645.6s
There's something I want to come back to. You talked about how when new models come out, you often have to kind of revisit things you've built. That's so interesting and so like frustrating maybe just like oh god damn it we shipped this thing now we have to rethink it. Talk about just like how
3646.5s
to. You talked about how when new models
3648.8s
come out, you often have to kind of
3650.2s
revisit things you've built. That's so
3652.2s
interesting and so like frustrating
3653.5s
maybe just like oh god damn it we
3655.4s
shipped this thing now we have to
3656.3s
rethink it. Talk about just like how
3657.9s
often you have to come back with a new model and we're like okay we have to redo this product that we launched a few months ago. A lot of the changes that we make with a new model is removing features that are no longer needed. So a lot of times we
3659.7s
model and we're like okay we have to
3661.2s
redo this product that we launched a few
3662.8s
months ago.
3663.8s
A lot of the changes that we make with a
3666.6s
new model is removing features that are
3670.3s
no longer needed. So a lot of times we
3673.6s
add features to the product as a crutch for the model because it's not naturally doing itself. So the classic example for this is a to-do list. When we first launched Quad Code, people would ask it to do these large refactors and Quad Code would say, Okay, cool. I need to change these like 20 call sites and it
3676.3s
for the model because it's not naturally
3678.2s
doing itself. So the classic example for
3680.8s
this is a to-do list. When we first
3683.0s
launched Quad Code, people would ask it
3684.8s
to do these large refactors and Quad
3686.6s
Code would say, Okay, cool. I need to
3688.2s
change these like 20 call sites and it
3690.3s
would go and change five of them and then stop. And then we were like, Okay, how do we like force it to remember to get every single one of these 20? And so Sid on our team was like, Okay, what if we just like think about what a human would do? So a human
3691.9s
then stop. And then we were like,
3694.1s
Okay, how do we like force it to
3696.4s
remember to get every single one of
3697.8s
these 20? And so Sid on our team was
3700.3s
like, Okay, what if we just like think
3702.5s
about what a human would do? So a human
3703.8s
would like make a list of everything that they need to change. Similar to how in VS Code you would look up all the call sites and it would be a list on the left side and you would like go through them one by one and replace all. How do we give this kind of like a tool to
3705.3s
that they need to change. Similar to how
3707.0s
in VS Code you would look up all the
3708.6s
call sites and it would be a list on the
3710.2s
left side and you would like go through
3711.7s
them one by one and replace all. How do
3713.6s
we give this kind of like a tool to
3715.0s
claude? And so he added a to-do list and we found that with that Claude was actually able to fix all these 20 call sites. But then with Opus 4 and later models we realized that we didn't need to force it to use this to-do list. It would like naturally use it itself. For
3717.9s
we found that with that Claude was
3719.9s
actually able to fix all these 20 call
3721.8s
sites. But then with Opus 4 and later
3724.2s
models we realized that we didn't need
3726.6s
to force it to use this to-do list. It
3730.1s
would like naturally use it itself. For
3732.3s
the earlier models, we had to keep reminding it, hey, did you finish everything on the to-do list? You can't finish until you're done with everything on the to-do list. And for the later models, without prompting, it just like naturally thinks to do everything on the to-do list. Um, these days, the to-do
3733.6s
reminding it, hey, did you finish
3735.1s
everything on the to-do list? You can't
3736.4s
finish until you're done with everything
3737.5s
on the to-do list. And for the later
3739.0s
models, without prompting, it just like
3741.1s
naturally thinks to do everything on the
3742.9s
to-do list. Um, these days, the to-do
3746.0s
list is still nice to have as like a user. Um, because then you can more clearly see what Claude is working on. But honestly, it's such a deemphasized part of the product right now that um, the model may use it, the model may not use it. it's like really not necessary for it to make thorough changes anymore.
3748.6s
user. Um, because then you can more
3751.1s
clearly see what Claude is working on.
3753.3s
But honestly, it's such a deemphasized
3755.2s
part of the product right now that um,
3757.6s
the model may use it, the model may not
3759.1s
use it. it's like really not necessary
3761.1s
for it to make thorough changes anymore.
3764.5s
I forget who said this on the podcast um that the model will eat your harness for breakfast. And what I'm hearing here is essentially you you remove things over time that you've had to add on top of the model where it was not operating the way you wanted. And essentially as the
3767.0s
that the model will eat your harness for
3768.7s
breakfast. And what I'm hearing here is
3770.7s
essentially you you remove things over
3773.4s
time that you've had to add on top of
3775.8s
the model where it was not operating the
3778.2s
way you wanted. And essentially as the
3779.5s
models get smarter, you just it becomes simpler and simpler for it just to do the thing you want it to do. Yeah. Um, we can move remove a lot of prompting interventions every time the model gets smarter. And we actually do this every time we launch a model. We read through the entire system prompt
3781.2s
simpler and simpler for it just to do
3782.6s
the thing you want it to do.
3784.1s
Yeah. Um, we can move remove a lot of
3786.6s
prompting interventions every time the
3788.8s
model gets smarter. And we actually do
3791.0s
this every time we launch a model. We
3792.6s
read through the entire system prompt
3794.4s
and we reflect on, okay, for each of these sections, does the model really need this reminder anymore? And if not, we'll remove it. The most exciting thing that new models unlocks though is just like entirely new features. So there's a lot of features that we've been testing out with prior models and the accuracy
3796.5s
these sections, does the model really
3798.2s
need this reminder anymore? And if not,
3800.4s
we'll remove it. The most exciting thing
3802.6s
that new models unlocks though is just
3804.3s
like entirely new features. So there's a
3807.8s
lot of features that we've been testing
3809.4s
out with prior models and the accuracy
3812.1s
wasn't high enough for us to want to launch them. And so one example of this is code review. We tried to build a code review product a few times and we've launched like simpler versions of code review which is the slashcode review command in the past and it was only with
3813.7s
launch them. And so one example of this
3816.1s
is code review. We tried to build a code
3818.8s
review product a few times and we've
3821.5s
launched like simpler versions of code
3823.6s
review which is the slashcode review
3825.4s
command in the past and it was only with
3827.6s
the most recent models that we felt like okay this code review is so good that our engineering team relies on this code review to pass before we merge PRs and we found that this was we've always dreamed of quad being able to be a reliable code reviewer that can actually that we can like confidently feel
3829.8s
okay this code review is so good that
3833.3s
our engineering team relies on this code
3835.6s
review to pass before we merge PRs and
3839.2s
we found that this was we've always
3842.2s
dreamed of quad being able to be a
3845.0s
reliable code reviewer that can actually
3847.4s
that we can like confidently feel
3849.4s
catches the majority of bugs. And it was only with like Opus 45 and 46 that we and uh Sonnet 4.6 that we felt like okay we are now able to like run multiple code review agents simultaneously to traverse traverse the entirety of the codebase and to synthesize a set of like real issues that an engineer needs to
3852.0s
only with like Opus 45 and 46 that we
3854.4s
and uh Sonnet 4.6 that we felt like okay
3857.4s
we are now able to like run multiple
3861.4s
code review agents simultaneously to
3864.1s
traverse traverse the entirety of the
3866.5s
codebase and to synthesize a set of like
3871.4s
real issues that an engineer needs to
3873.4s
address before merge. And so this is like a new capability that the the newest models have unlocked. This is another trend that is very common on this podcast of build something that will possibly be possible in the next six months. Be kind of at the edge of what's working sort of and
3875.1s
like a new capability that the the
3877.4s
newest models have unlocked.
3879.2s
This is another trend that is very
3881.0s
common on this podcast of build
3882.9s
something that will possibly be possible
3885.4s
in the next six months. Be kind of at
3887.2s
the edge of what's working sort of and
3889.5s
then it'll catch up and then it'll be an amazing product and you'll be ahead of amazing product and you'll be ahead of everyone. Yeah, exactly. Um it's pretty important to build products that don't necessarily work yet so that you know okay what is missing um for this product to work and
3891.0s
amazing product and you'll be ahead of
3892.4s
amazing product and you'll be ahead of everyone.
3893.0s
Yeah, exactly. Um it's pretty important
3895.1s
to build products that don't necessarily
3898.0s
work yet so that you know okay what is
3901.4s
missing um for this product to work and
3904.5s
then with the newest model you can just swap it in to the prototype you've already made and see okay does this new model close that gap. How much are you able to speak to just kind of where things are going with claude and co-work as kind of the vision of it? I imagine you don't want to give
3906.7s
swap it in to the prototype you've
3908.7s
already made and see okay does this new
3911.0s
model close that gap.
3912.2s
How much are you able to speak to just
3913.8s
kind of where things are going with
3915.9s
claude and co-work as kind of the vision
3917.8s
of it? I imagine you don't want to give
3919.5s
away too much about the goal but it feels like you're there's all these awesome features being added on top dispatch control from phone and all these mobile app all these things what's kind of just like a way to understand the vision for all these things long the vision for all these things long term
3921.3s
feels like you're there's all these
3922.3s
awesome features being added on top
3924.0s
dispatch control from phone and all
3926.6s
these mobile app all these things what's
3928.5s
kind of just like a way to understand
3930.4s
the vision for all these things long
3932.2s
the vision for all these things long term
3932.8s
we think about this in terms of building blocks so for both quad code and co-work the core building block is making individual tasks successful so you you want to produce some output you give it a clear prompt description is it able to consistently produce acceptable output that you're able to either merge or
3934.2s
blocks so for both quad code and co-work
3937.7s
the core building block is making
3940.1s
individual tasks successful so you you
3943.6s
want to produce some output you give it
3946.1s
a clear prompt description is it able to
3948.6s
consistently produce acceptable output
3951.1s
that you're able to either merge or
3953.3s
share with your colleagues or external audience. So the task is the core building block. As the models get smarter, the task success rate gets a lot higher. And then we see people moving towards doing multiple tasks at the same time. So multi-coding was this big thing in towards the end of 2025 and
3954.7s
audience. So the task is the core
3956.5s
building block. As the models get
3958.4s
smarter, the task success rate gets a
3960.6s
lot higher. And then we see people
3962.6s
moving towards doing multiple tasks at
3964.8s
the same time. So multi-coding was this
3966.8s
big thing in towards the end of 2025 and
3969.3s
it's only increased since then. And so we see this as okay great one task works and now you can do like six tasks at a time. As the models get even smarter the way that we are extrapolating this is okay next maybe you're going to run like 50 clouds at a time or hundreds of
3971.5s
we see this as okay great one task works
3975.0s
and now you can do like six tasks at a
3977.2s
time. As the models get even smarter the
3980.1s
way that we are extrapolating this is
3981.8s
okay next maybe you're going to run like
3985.0s
50 clouds at a time or hundreds of
3986.7s
clouds at a time. And so what is the infrastructure we need to build to enable that? At that point you're probably not going to run everything locally on your machine anymore. There's just like not enough RAM to do it. And so we're we're thinking about h how do we make it easier for you to manage all
3988.6s
infrastructure we need to build to
3989.7s
enable that? At that point you're
3992.0s
probably not going to run everything
3993.5s
locally on your machine anymore. There's
3995.3s
just like not enough RAM to do it. And
3999.1s
so we're we're thinking about h how do
4002.6s
we make it easier for you to manage all
4004.3s
these? These will probably run remotely. How do we build the interface so that you as a human know which tasks you need to look look into? How do we make sure that the agent is fully verifying work so that when you look at a task and it says it's done, you like can very
4006.6s
How do we build the interface so that
4008.4s
you as a human know which tasks you need
4011.4s
to look look into? How do we make sure
4014.2s
that the agent is fully verifying work
4016.1s
so that when you look at a task and it
4018.7s
says it's done, you like can very
4021.2s
quickly verify and fully trust that it is done to your spec. and how do we make sure that this like process is self-improving so that when you do see a task that isn't done to your liking, you can give it feedback and the model will know for every future run to incorporate
4023.0s
is done to your spec. and how do we make
4025.7s
sure that this like process is
4027.0s
self-improving so that when you do see a
4029.8s
task that isn't done to your liking, you
4032.6s
can give it feedback and the model will
4034.7s
know for every future run to incorporate
4036.7s
that feedback so it never makes that mistake again. So this is the progression that we're we're bringing our users along for. There's a lot of people listening, a lot of product managers, a lot of maybe founders, a lot of other cross functional folks listening. There's a lot of worry about just how their role
4038.1s
mistake again. So this is the
4040.3s
progression that we're we're bringing
4041.8s
our users along for.
4043.2s
There's a lot of people listening, a lot
4044.4s
of product managers, a lot of maybe
4046.5s
founders, a lot of other cross
4048.3s
functional folks listening. There's a
4050.6s
lot of worry about just how their role
4053.6s
just the future of their careers. What advice would you have for just people to not just survive this transition to this very AIdriven world, but to be really successful to essentially just to thrive in this future? What are just like things people need to hear, need to be things people need to hear, need to be doing?
4056.2s
advice would you have for just people to
4059.2s
not just survive this transition to this
4061.6s
very AIdriven world, but to be really
4063.9s
successful to essentially just to thrive
4065.7s
in this future? What are just like
4067.2s
things people need to hear, need to be
4069.0s
things people need to hear, need to be doing?
4069.7s
I think AI gives everybody a ton more leverage than they used to. And so I would push you towards anytime you realize that you're doing some manual task multiple times, think about how you can use cloud code, co-work or other AI tools to automate that for you. Most people have like creative parts of their
4072.6s
leverage than they used to. And so I
4075.3s
would push you towards anytime you
4077.7s
realize that you're doing some manual
4079.3s
task multiple times, think about how you
4081.8s
can use cloud code, co-work or other AI
4084.2s
tools to automate that for you. Most
4087.6s
people have like creative parts of their
4090.4s
job that they absolutely love and then like tedious parts of their job that they really hate doing. I think the beauty of AI is that it can do those tedious parts for you. it can learn from every time that you've done that manual task and generalize and then run it automatically and so that you can focus
4092.5s
like tedious parts of their job that
4094.3s
they really hate doing. I think the
4096.3s
beauty of AI is that it can do those
4099.3s
tedious parts for you. it can learn from
4101.8s
every time that you've done that manual
4103.7s
task and generalize and then run it
4106.2s
automatically and so that you can focus
4108.3s
on the creative parts and that means you can do a lot more than you used to be able to do. So I think my like immediate push for people is figure out the repetitive parts that you can pass to quad. Iterate on those automations until the success rate is very high and then
4110.6s
can do a lot more than you used to be
4111.8s
able to do. So I think my like immediate
4114.9s
push for people is figure out the
4116.8s
repetitive parts that you can pass to
4118.6s
quad. Iterate on those automations until
4120.9s
the success rate is very high and then
4123.4s
focus on okay what more can you be doing for your team for your product for your company that like people haven't had the bandwidth to pick up so far or like what is that like pet project that you always thought the company should do that like you've never had bandwidth to do. If AI
4126.2s
for your team for your product for your
4128.4s
company that like people haven't had the
4130.8s
bandwidth to pick up so far or like what
4133.4s
is that like pet project that you always
4135.3s
thought the company should do that like
4137.4s
you've never had bandwidth to do. If AI
4140.7s
can take care of the like grunt work, then you have you have this extra 20 time now that you might not have before. So, so my push is to lean into these tools, hand off the work that you're not excited to do, figure out how it can accelerate you, and then as a result,
4143.7s
then you have you have this extra 20
4146.1s
time now that you might not have before.
4148.9s
So, so my push is to lean into these
4151.8s
tools, hand off the work that you're not
4153.8s
excited to do, figure out how it can
4155.5s
accelerate you, and then as a result,
4157.7s
you'll be able to do so much more. Something core to what you just shared, which I fully agree with, is find problems to solve with AI. There's all this potential what all these tools can do. some of the hard like for a lot of people hardest part is just like what should I actually do and what you're
4159.4s
Something core to what you just shared,
4161.0s
which I fully agree with, is find
4162.9s
problems to solve with AI. There's all
4166.1s
this potential what all these tools can
4168.2s
do. some of the hard like for a lot of
4170.2s
people hardest part is just like what
4171.4s
should I actually do and what you're
4173.0s
saying here is just pay attention to things that you are doing constantly you can automate pay attention to just like ideas that have been floating around that you haven't had time to do um it's basically it's like solve a problem for yourself is kind of the core advice yourself is kind of the core advice there
4174.8s
things that you are doing constantly you
4176.4s
can automate pay attention to just like
4178.4s
ideas that have been floating around
4179.8s
that you haven't had time to do um it's
4182.1s
basically it's like solve a problem for
4183.4s
yourself is kind of the core advice
4185.0s
yourself is kind of the core advice there
4185.5s
exactly I I would also push listeners towards focusing on bringing your automations from okay this is a cool concept to like hey this actually works 100 of the time like sometimes I see users trying trying to automate something, getting it to like 90 95 accuracy and then giving up on it. And
4187.6s
towards focusing on bringing your
4190.9s
automations from okay this is a cool
4193.2s
concept to like hey this actually works
4194.9s
100 of the time like sometimes I see
4197.1s
users trying trying to automate
4199.0s
something, getting it to like 90 95
4201.8s
accuracy and then giving up on it. And
4205.5s
this if an automation doesn't work 100 of the time, it's not really an automation. And that last 5 to 10 does take more time. Also, building the automation is often a lot slower than you doing it yourself. I would encourage listeners to put in that time to scope some automation that you really want to
4208.5s
of the time, it's not really an
4209.8s
automation. And that last 5 to 10 does
4213.1s
take more time. Also, building the
4216.4s
automation is often a lot slower than
4218.6s
you doing it yourself. I would encourage
4221.4s
listeners to put in that time to scope
4225.0s
some automation that you really want to
4226.6s
get to 100. Put in the elbow grease to teach quality your preferences to like give it feedback so that it can improve its skill so that it can get to that 100. And then like really then you'll be able to rely on it. There there's just not much value in a 95 there
4230.8s
teach quality your preferences to like
4232.8s
give it feedback so that it can improve
4235.0s
its skill so that it can get to that
4237.0s
100. And then like really then you'll
4239.7s
be able to rely on it. There there's
4241.4s
just not much value in a 95 there
4243.6s
just not much value in a 95 there automation. I am super guilty of that. This is really good advice for me. I am guilty of this too. I've been teaching it I've been teaching co-work to try to get me to inbox zero for Gmail and it has not been it it has been very
4244.7s
I am super guilty of that. This is
4246.6s
really good advice for me.
4248.5s
I am guilty of this too. I've been
4250.0s
teaching it I've been teaching co-work
4252.0s
to try to get me to inbox zero for Gmail
4255.1s
and it has not been it it has been very
4258.6s
time consuming and it is definitely not there as you probably realized. Yeah, I funny enough that's exactly where my mind goes. I have this uh workflow I set up where every email I get, it looks for things that are spammy, which is just like all these like, Hey, can I come on your podcast
4260.6s
there as you probably realized.
4262.4s
Yeah, I funny enough that's exactly
4263.9s
where my mind goes. I have this uh
4266.0s
workflow I set up where every email I
4268.0s
get, it looks for things that are
4270.0s
spammy, which is just like all these
4271.9s
like, Hey, can I come on your podcast
4273.4s
or what about this one? Like all these things I'm just like, I don't have time for these sorts of things. And I have it categorized it into a folder called spammy. And it's just like it's 95 great, but then there's like, oh wow, I missed an email because it went in
4275.3s
things I'm just like, I don't have time
4276.6s
for these sorts of things. And I have it
4278.6s
categorized it into a folder called
4280.2s
spammy. And it's just like it's 95
4282.9s
great, but then there's like, oh wow, I
4284.4s
missed an email because it went in
4285.4s
there. So this is a good push for me to like I'm going to work on this. I'm going to get it to perfect. Yeah. We also are working on making the flow for customizing these commands a lot easier because right now I think you have to like know too many concepts. You
4287.0s
like I'm going to work on this. I'm
4288.1s
going to get it to perfect.
4289.0s
Yeah. We also are working on making the
4291.6s
flow for customizing these commands a
4294.2s
lot easier because right now I think you
4295.5s
have to like know too many concepts. You
4297.4s
have to know to define a skill. You have to know to like use this skill and give it feedback. And then you have to know to tell co-work to update the skill based on all the feedback that you gave. And then you also have to know where to read the skill to like make sure that
4299.2s
to know to like use this skill and give
4300.9s
it feedback. And then you have to know
4302.6s
to tell co-work to update the skill
4305.3s
based on all the feedback that you gave.
4307.4s
And then you also have to know where to
4309.2s
read the skill to like make sure that
4310.6s
the feedback was incorporated the way that you want. The it's also our job to make this flow really seamless so that it doesn't feel painful to do. Amazing. Is there anything else, Cat, you wanted to share? Anything else you wanted to leave listeners with? Anything you wanted to double down on that we
4311.9s
that you want. The it's also our job to
4314.6s
make this flow really seamless so that
4316.4s
it doesn't feel painful to do.
4317.9s
Amazing. Is there anything else, Cat,
4319.9s
you wanted to share? Anything else you
4321.5s
wanted to leave listeners with? Anything
4323.1s
you wanted to double down on that we
4325.2s
haven't already touched on before we get to our very exciting lightning round? I see a lot of people playing around with AI um and building like prototype apps and tinkering with building workflows. I would really push people towards building apps that you're actually using every single day because I think only through that usage are you actually
4326.8s
to our very exciting lightning round? I
4328.6s
see a lot of people playing around with
4330.2s
AI um and building like prototype apps
4333.9s
and tinkering with building workflows. I
4337.6s
would really push people towards
4340.0s
building apps that you're actually using
4341.8s
every single day because I think only
4343.5s
through that usage are you actually
4345.3s
getting the value. Like if you build a prototype app that isn't helping you get more done, then the the AI isn't really adding value to the the AI isn't really adding value to your to your day. And there's only so much you learn from that when it's like, okay, I just one-shoted something. Oh, that's cool.
4347.1s
prototype app that
4350.0s
isn't helping you get more done, then
4353.7s
the the AI isn't really adding value to
4356.2s
the the AI isn't really adding value to your
4357.0s
to your day.
4358.1s
And there's only so much you learn from
4359.3s
that when it's like, okay, I just
4361.5s
one-shoted something. Oh, that's cool.
4363.0s
And then you never come back to it. Like you're not learning a lot and you're not getting like much leverage from it and actual leverage. Yeah, that's such a good point. I also think there's a lot of people who spend a lot of time like customizing their workflow. So there's like I think
4364.2s
you're not learning a lot
4365.1s
and you're not getting like much
4366.5s
leverage from it
4367.2s
and actual leverage. Yeah, that's such a
4368.6s
good point.
4369.1s
I also think there's a lot of people who
4370.5s
spend a lot of time like customizing
4372.9s
their workflow. So there's like I think
4374.6s
there's like two ends of the spectrum. One is like people who never customize or never build automations, but there's like this polar opposite end of people who like obsess around customizing their tool like adding a ton of skills and MCPs and um these like workflow improvements and I think sometimes that
4375.8s
One is like people who never customize
4377.3s
or never build automations, but there's
4379.2s
like this polar opposite end of people
4381.0s
who like obsess around customizing their
4384.6s
tool like adding a ton of skills and
4387.9s
MCPs and um these like workflow
4391.0s
improvements and I think sometimes that
4393.3s
can even distract from your core goal of like launching some product or building some feature. I think there's a lot of fun in customizing and we definitely want to make our products very hackable so that you you can make it work really well for you, but there is a limit to
4395.7s
like launching some product or building
4397.6s
some feature. I think there's a lot of
4399.4s
fun in customizing and we definitely
4401.6s
want to make our products very hackable
4403.6s
so that you you can make it work really
4405.9s
well for you, but there is a limit to
4409.2s
how much it's useful. Um, and I think there there's a camp of people who maybe spend so much time customizing that they're like not sleeping and not doing the like core task that they originally set out to do. I see a lot of that on Twitter just like look at my setup. It's out of control.
4412.3s
there there's a camp of people who maybe
4414.6s
spend so much time customizing that
4416.4s
they're like not sleeping and not doing
4418.5s
the like core task that they originally
4420.6s
set out to do.
4421.4s
I see a lot of that on Twitter just like
4424.2s
look at my setup. It's out of control.
4425.8s
It's so optimized. Then what are you what what are you actually building? No, but my setup is so awesome. Like it gets so much done. I think the simple setups actually work I think the simple setups actually work better. Sl powerup getting take level up a little bit. Yeah. Yeah. There's this Karpathy tweet that just uh
4427.5s
what what are you actually building? No,
4429.8s
but my setup is so awesome. Like it gets
4431.8s
so much done.
4432.8s
I think the simple setups actually work
4434.8s
I think the simple setups actually work better.
4436.3s
Sl powerup getting take level up a
4438.5s
little bit.
4438.9s
Yeah. Yeah.
4439.7s
There's this Karpathy tweet that just uh
4442.2s
came out yesterday where he talked about this divide that's interesting between people that tried chatbt claw back in the day. it was like okay and they're like nah this is this is terrible and they kind of gave up on like what AI could do for them and they're just like so cynical of like no way it's not
4443.8s
this divide that's interesting between
4446.2s
people that tried chatbt claw back in
4449.4s
the day. it was like okay and they're
4451.0s
like nah this is this is terrible and
4452.6s
they kind of gave up on like what AI
4454.5s
could do for them and they're just like
4455.5s
so cynical of like no way it's not
4457.0s
actually that big of a deal and then there's people that are using it to code essentially who see the full intense power of it and how good it is and people on both sides don't understand the other side and why they like how much they how they see the world and so
4458.5s
there's people that are using it to code
4460.9s
essentially who see the full intense
4464.0s
power of it and how good it is and
4467.1s
people on both sides don't understand
4468.7s
the other side and why they like how
4470.6s
much they how they see the world and so
4472.9s
your advice is really good here just like actually use it for real things and see how good it actually has gotten yeah I think The big shift is that the 2024 generation of products were chatbased and the quad code generation of products is action-based. of products is action-based. And the like big aha moment people have is
4474.3s
like actually use it for real things and
4476.0s
see how good it actually has gotten
4478.4s
yeah I think The big shift is that the
4480.9s
2024 generation of products were
4483.3s
chatbased and the quad code generation
4486.6s
of products is action-based.
4489.3s
of products is action-based. And
4490.9s
the like big aha moment people have is
4493.0s
when quad can just like do things on your behalf. It is it is an amazing feeling to know that the agent is capable of doing so much more than telling you what to do. Like the agent can actually just do it itself. And when people feel that, I I think that's the eye opening moment.
4495.0s
your behalf. It is it is an amazing
4497.8s
feeling to know that the agent is
4500.5s
capable of doing so much more than
4502.5s
telling you what to do. Like the agent
4504.5s
can actually just do it itself. And when
4507.0s
people feel that, I I think that's the
4509.0s
eye opening moment.
4510.5s
Shout out uh Chrome extension, the cloud called Chrome extension, which you can just watch it doing stuff and you'd be like, Fill out this form for me and like, all right, here I go. like, all right, here I go. Exactly. Okay. Uh anything else before we get to our very exciting lightning round? No, let's do it.
4512.6s
called Chrome extension, which you can
4513.8s
just watch it doing stuff and you'd be
4515.3s
like, Fill out this form for me and
4516.9s
like, all right, here I go.
4518.2s
like, all right, here I go. Exactly.
4519.2s
Okay. Uh anything else before we get to
4521.0s
our very exciting lightning round?
4522.8s
No, let's do it.
4524.0s
Let's do it. Uh Kat, I've got five questions for you. Welcome to the lightning round. There's this animation that place. I have to make sure to say it. Uh are you ready? I'm ready. First question, what are two or three books that you find yourself recommending most to other people?
4526.3s
questions for you. Welcome to the
4528.1s
lightning round. There's this animation
4529.3s
that place. I have to make sure to say
4530.7s
it. Uh are you ready?
4532.6s
I'm ready.
4534.0s
First question, what are two or three
4536.0s
books that you find yourself
4537.1s
recommending most to other people?
4538.7s
I really like how Asia works. Um, it's a story about economic development and what are like the policies and uh governments that make um long lasting successful economies. The other books that I'm really into are the technology trap. So, this is actually about the past few technology revolutions. So the industrial revolution and the computer
4542.4s
story about economic development and
4545.5s
what are like the policies and uh
4548.6s
governments that make um long lasting
4551.4s
successful economies. The other books
4554.2s
that I'm really into are the technology
4556.3s
trap. So, this is actually about the
4559.1s
past few technology revolutions. So the
4561.8s
industrial revolution and the computer
4563.2s
revolution and how this has affected uh workers. The the reason that I really like this is because I think we there's a lot we can learn from history to make sure that this transition goes well. And um maybe on like a fun note, I really like paper menagerie. Um it's just like
4566.4s
workers. The the reason that I really
4569.0s
like this is because I think we there's
4571.2s
a lot we can learn from history to make
4573.4s
sure that this transition goes well. And
4577.1s
um maybe on like a fun note, I really
4579.0s
like paper menagerie. Um it's just like
4581.6s
a book of short stories about like coming of age and AI and um just like self-discovery. Favorite recent movie or TV show you have really enjoyed? I really like Drive to Survive. There's no like deeper meaning to it. I There's no like deeper meaning to it. I just there's just something very satisfying
4584.2s
coming of age and AI and um just like
4589.7s
self-discovery. Favorite recent movie or
4592.3s
TV show you have really enjoyed?
4594.0s
I really like Drive to Survive.
4597.5s
There's no like deeper meaning to it. I
4599.4s
There's no like deeper meaning to it. I just
4601.2s
there's just something very satisfying
4602.9s
about people being so obsessed with like a singular engineering goal and just like the purity of their pursuit. Um, and I also really love Free Solo, which is about Alex Honold um, climbing El Capetan without a harness. And I think similarly, it's just such a pure achievement to be able to climb this extremely challenging, dangerous route
4606.2s
a singular engineering goal and just
4609.9s
like the purity of their pursuit. Um,
4612.2s
and I also really love Free Solo, which
4614.2s
is about Alex Honold um, climbing El
4618.4s
Capetan without a harness. And I think
4621.5s
similarly, it's just such a pure
4624.5s
achievement to be able to climb this
4627.7s
extremely challenging, dangerous route
4631.0s
and to be able to have the mental focus to do it knowing that if you make a single mistake, you die. It's insane. Yeah, that movie is out of control. And it's interesting how these relate in some way to the work you do. I actually am a rock climber. Um I first
4633.4s
to do it knowing that if you make a
4636.2s
single mistake, you die.
4637.5s
It's insane. Yeah, that movie is out of
4639.0s
control. And it's interesting how these
4640.6s
relate in some way to the work you do.
4642.4s
I actually am a rock climber. Um I first
4645.0s
watched Free Solo before I climbed rocks and so I thought it was impressive. I didn't understand how impressive it was. It's one of the rare movies where like the more you know about it, the more you're you're blown away by how insane this is. Like the kinds the kinds of moves he's doing on the wall are things
4647.4s
and so I thought it was impressive. I
4649.5s
didn't understand how impressive it was.
4651.0s
It's one of the rare movies where like
4652.5s
the more you know about it, the more
4653.9s
you're you're blown away by how insane
4656.6s
this is. Like the kinds the kinds of
4659.4s
moves he's doing on the wall are things
4660.9s
that like I don't think I will ever be able to do in my lifetime if it were set in a gym like one ft off the ground with a rope. With a rope. Did you see the documentary on that other guy, the younger one that went on like ice mountain? I did. That one was very sad.
4662.8s
able to do in my lifetime if it were set
4665.0s
in a gym like one ft off the ground
4667.8s
with a rope.
4668.8s
With a rope.
4670.3s
Did you see the documentary on that
4671.8s
other guy, the younger one that went on
4673.6s
like ice mountain?
4674.7s
I did. That one was very sad.
4676.2s
But that was that was wild. Okay. Uh favorite product you recently discovered that you really love? The product that is like most changed my life outside of cloud products is probably Whimo. Like I'm a diehard Whimo user. Um use it twice a day, get to and from work. So the two things that I really like about
4678.3s
favorite product you recently discovered
4679.8s
that you really love? The product that
4681.8s
is like most changed my life outside of
4684.7s
cloud products is probably Whimo.
4687.7s
Like I'm a diehard Whimo user. Um use it
4690.8s
twice a day, get to and from work. So
4693.0s
the two things that I really like about
4694.2s
it are one, I don't feel bad if a Whimo is waiting for me. And so I feel like I feel less pressure to be right at the curbside the moment it arrives. And the second thing is I feel like it lets me be a bit more productive. Um, when when
4697.6s
is waiting for me. And so I feel like I
4700.1s
feel less pressure to be right at the
4703.0s
curbside the moment it arrives. And the
4705.8s
second thing is I feel like it lets me
4708.0s
be a bit more productive. Um, when when
4711.0s
I'm in the car with another human, I I typically try not to like do any work calls. I I feel a little rude if I'm like on my laptop the whole time. But one thing I really appreciate about the Whimo is I can call into a work call. I'm not worried about someone
4713.3s
typically try not to like do any work
4715.2s
calls. I I feel a little rude if I'm
4717.4s
like on my laptop the whole time. But
4719.2s
one thing I really appreciate about the
4720.6s
Whimo is I can call into a work call.
4723.0s
I'm not worried about someone
4724.2s
overhearing me. I'm not worried about, hey, is this like rude? Am I talking too loud? Do I need to tell ask someone to like change the music? And so this has been like I feel like this has given me back like 30 minutes every day. All these second order effects of of technology. It's so interesting.
4726.8s
hey, is this like rude? Am I talking too
4728.5s
loud? Do I need to tell ask someone to
4730.2s
like change the music? And so this has
4731.9s
been like I feel like this has given me
4733.8s
back like 30 minutes every day.
4735.6s
All these second order effects of of
4737.5s
technology. It's so interesting.
4739.1s
Yeah. I always thought Whimo needed to be priced lower than Uber and Lyft to succeed, but actually I'm like very happy to pay a 2x premium for it. I love Whimo. It's just like like once you see it, you're just like, Wow, this is insane. And and then you get used to
4740.8s
be priced lower than Uber and Lyft to
4743.0s
succeed, but actually I'm like very
4745.0s
happy to pay a 2x premium for it.
4747.0s
I love Whimo. It's just like like once
4749.3s
you see it, you're just like, Wow, this
4751.5s
is insane. And and then you get used to
4753.5s
it. Like you get in there, you're like, This is crazy. And then you forget about it. Totally. And I think it's also changed the vernacular. Like a lot of people at Anthropic love Whimo. And I think in the past you would be like, Hey, like let's call like blah blah ride share app. And
4755.0s
This is crazy. And then you forget
4757.1s
about it.
4757.7s
Totally. And I think it's also changed
4759.0s
the vernacular. Like a lot of people at
4761.2s
Anthropic love Whimo. And I think in the
4764.0s
past you would be like, Hey, like let's
4766.3s
call like blah blah ride share app. And
4768.0s
now like everyone's just like, Okay, is the way here? Okay, two more questions. Do you have a favorite life motto that you often come back to in work or in life? Just do things. That's right. I think there's a lot of value in like first principles thinking and if if you like if you know what you're optimizing
4769.4s
the way here?
4770.4s
Okay, two more questions. Do you have a
4772.4s
favorite life motto that you often come
4773.9s
back to in work or in life?
4775.4s
Just do things.
4777.8s
That's right.
4778.4s
I think there's a lot of value in like
4780.1s
first principles thinking and if if you
4783.3s
like if you know what you're optimizing
4784.9s
for and you have like strong first principles, then you can normally deduce what the right like course of action is and be able to clearly articulate that to all the stakeholders and then you should just like do it. Like I think jobs are fake. If you understand the constraints, you can figure out what you
4786.2s
principles, then you can normally deduce
4788.0s
what the right like course of action is
4789.9s
and be able to clearly articulate that
4791.8s
to all the stakeholders and then you
4794.6s
should just like do it. Like I think
4796.4s
jobs are fake. If you understand the
4798.7s
constraints, you can figure out what you
4801.2s
can do and then just like try to do it quickly, learn from the mistakes and apologize or fix them if you did something wrong. You you could just do things whoever said that. I think it's liberating actually to like tell people this. I think in a lot of companies like roles are very strictly
4803.0s
quickly, learn from the mistakes and
4805.6s
apologize or fix them if you did
4807.1s
something wrong.
4808.0s
You you could just do things whoever
4810.0s
said that.
4810.5s
I think it's liberating actually to like
4812.2s
tell people this. I think in a lot of
4813.9s
companies like roles are very strictly
4816.9s
defined like okay this is what the PM does, this is what the designer does, this is what engineer does and then even team scopes are very rigidly defined. So, hey, like this corner of the codebase we touch and this corner like we're not allowed to touch. And I think what just do things lets people do is
4819.5s
does, this is what the designer does,
4820.7s
this is what engineer does and then even
4822.8s
team scopes are very rigidly defined.
4824.9s
So, hey, like this corner of the
4826.3s
codebase we touch and this corner like
4828.2s
we're not allowed to touch. And I think
4830.4s
what just do things lets people do is
4832.2s
they feel like empowered to make these decisions, empowered to operate across team boundaries just to like get something done. That feels like a big important skill to be good at. People call it agency. Just like do the things bias towards action. All these ways of describing just like you wait for describing just like you wait for permission.
4833.8s
decisions, empowered to operate across
4836.4s
team boundaries just to like get
4837.9s
something done.
4838.5s
That feels like a big important skill to
4841.8s
be good at. People call it agency. Just
4844.6s
like do the things
4846.8s
bias towards action. All these ways of
4848.3s
describing just like you wait for
4850.0s
describing just like you wait for permission.
4851.0s
Yeah. I think this is my favorite reason to work at a startup at some point in your life because like one thing that was like very life-changing for me was actually working at scale when we were 20 people. And so there was just no process and we had like really big
4852.9s
to work at a startup at some point in
4854.6s
your life because like one thing that
4856.7s
was like very life-changing for me was
4858.6s
actually working at scale when we were
4860.1s
20 people. And so there was just no
4862.8s
process and we had like really big
4864.2s
problems that we needed to solve. And it it was like I really appreciate Alex and the rest of the team for like empowering me and the rest of the team to just like figure things out without any boundaries for what sales supposed to do, what off supposed to do, what engineer is
4867.0s
it was like I really appreciate Alex and
4869.6s
the rest of the team for like empowering
4871.0s
me and the rest of the team to just like
4873.9s
figure things out without any boundaries
4875.8s
for what sales supposed to do, what off
4877.4s
supposed to do, what engineer is
4878.4s
supposed to do. just like you have all the tools at your disposal. You have some like ambitious hairy problem statement and you can do whatever you need to like get to a good solution. Like you almost need that experience to build that skill to feel comfortable doing that because a lot of people, you
4880.6s
the tools at your disposal. You have
4883.4s
some like ambitious hairy problem
4885.0s
statement and you can do whatever you
4886.6s
need to like get to a good solution.
4888.6s
Like you almost need that experience to
4890.2s
build that skill to feel comfortable
4891.7s
doing that because a lot of people, you
4893.3s
know, they go through school or in college and all these like do the thing we tell you to do and then you will get a good grade. And you have to kind of unlearn that of like, okay, I'm just going to do the thing that needs to be done and even if people think it's dumb,
4894.8s
college and all these like do the thing
4896.6s
we tell you to do and then you will get
4898.2s
a good grade. And you have to kind of
4899.8s
unlearn that of like, okay, I'm just
4901.2s
going to do the thing that needs to be
4902.2s
done and even if people think it's dumb,
4904.6s
I think it's the right thing to do. Yeah. Exactly. Okay. Okay, I actually have two more quick questions. Two more final questions. One is uh when Claude thinks, there's all these I don't know if you call them verbs. What's the term for these things? Uh thinking words. Thinking words. And interestingly, these
4906.3s
Yeah. Exactly.
4907.5s
Okay. Okay, I actually have two more
4908.6s
quick questions. Two more final
4909.7s
questions. One is uh when Claude thinks,
4911.8s
there's all these I don't know if you
4913.0s
call them verbs. What's the term for
4914.7s
these things?
4915.4s
Uh thinking words.
4916.7s
Thinking words. And interestingly, these
4918.4s
all leaked in the source code. Uh is it Do you have a favorite thinking word? I really like manifesting. It's also like the sticker that I I have on my like the sticker that I I have on my favorite. Clearly the winner. Okay, final question. Asked Boris this too. with AGI potentially arriving in our lifetime
4921.0s
Do you have a favorite thinking word?
4923.7s
I really like manifesting. It's also
4926.1s
like the sticker that I I have on my
4930.2s
like the sticker that I I have on my favorite.
4930.6s
Clearly the winner. Okay, final
4932.2s
question. Asked Boris this too. with AGI
4935.9s
potentially arriving in our lifetime
4938.6s
when you don't potentially have to work, what are you going to do? What are you going to do with all your time? I think it it will take a long time for AGI to diffuse across society. So, I think the immediate thing is actually just like helping bring the world along. I think my like non-serious answer for
4941.4s
what are you going to do? What are you
4942.5s
going to do with all your time?
4943.7s
I think it it will take a long time for
4946.1s
AGI to diffuse across society. So, I
4949.0s
think the immediate thing is actually
4950.4s
just like helping bring the world along.
4953.0s
I think my like non-serious answer for
4955.0s
after this happens is I'll probably just do a lot of rock climbing. I'll probably just like live in some I'll probably move to like Fountain Blue and just like live amongst 10,000 boulders and climb for a bit. There's also so many books I want to read that my my goal is to be
4958.5s
do a lot of rock climbing. I'll probably
4960.2s
just like live in some I'll probably
4962.5s
move to like Fountain Blue and just like
4964.6s
live amongst 10,000 boulders and climb
4968.1s
for a bit. There's also so many books I
4970.6s
want to read that my my goal is to be
4973.9s
able to read one or two books a week and I'm currently at probably like 0.5. there's just like so much we can learn from history and so much that I don't understand as well as I would love to. Like I don't know anything about physics and or like robotics or like any hardware or like aerospace or there's
4977.4s
I'm currently at probably like 0.5.
4984.1s
there's just like so much we can learn
4985.3s
from history and so much that I don't
4987.7s
understand as well as I would love to.
4989.8s
Like I don't know anything about physics
4991.5s
and or like robotics or like any
4995.6s
hardware or like aerospace or there's
4998.3s
just so many interesting topics. So I I'm excited to learn even even knowing that the AI will already know it. Cat, this was amazing. You're awesome. Two follow questions. Where can folks find you online if they want to reach out and just follow what you're up to? And how can listeners be useful to you?
5000.7s
I'm excited to learn even even knowing
5002.8s
that the AI will already know it.
5006.2s
Cat, this was amazing. You're awesome.
5009.4s
Two follow questions. Where can folks
5010.8s
find you online if they want to reach
5012.1s
out and just follow what you're up to?
5013.5s
And how can listeners be useful to you?
5015.4s
The best way to reach out is I am Catwoo on Twitter. Um, feel free to like tag me in things. Feel free to DM me. I read all all my DMs. I don't always respond to every single one, but I will read them all. And then the thing that is most helpful is tell us where cloud code
5019.3s
on Twitter. Um, feel free to like tag me
5023.5s
in things. Feel free to DM me. I read
5026.6s
all all my DMs. I don't always respond
5028.4s
to every single one, but I will read
5029.8s
them all. And then the thing that is
5033.1s
most helpful is tell us where cloud code
5036.4s
and co-work aren't working well for you. We we are very grateful for the amount of positive feedback. But the things that we thrive on is edge cases, errors, like specific tasks that we can reproduce where quad code or co-work fail. Because if you're able to share that with us and we're able to reproduce
5039.0s
We we are very grateful for the amount
5041.4s
of positive feedback. But the things
5044.0s
that we thrive on is edge cases, errors,
5048.7s
like specific tasks that we can
5050.8s
reproduce where quad code or co-work
5053.9s
fail. Because if you're able to share
5056.4s
that with us and we're able to reproduce
5058.1s
it, then this is something that we're able to actively improve for our next generations of models and uh for our next harnesses. Extremely cool. everyone on people on Twitter are not shy with sharing this feedback. So, so keep it coming. Share us share, please, please share the problems that you're having with us.
5059.8s
able to actively improve for our next
5061.5s
generations of models and uh for our
5064.1s
next harnesses.
5065.7s
Extremely cool. everyone on people on
5067.4s
Twitter are not shy with sharing this
5068.8s
feedback. So, so keep it coming.
5070.9s
Share us share, please, please share the
5073.2s
problems that you're having with us.
5074.4s
Yeah. And it's really cool to see all you your team being on so active on Twitter and responding to people and so so like what I'm hearing like this is actually stuff you guys actually see and react to. So yeah, we appreciate everyone being so engaged with us. Um it gives the team a
5075.6s
you your team being on so active on
5077.4s
Twitter and responding to people and so
5080.4s
so like what I'm hearing like this is
5081.9s
actually stuff you guys actually see and
5083.4s
react to. So
5084.4s
yeah, we appreciate everyone being so
5085.9s
engaged with us. Um it gives the team a
5088.0s
ton of energy. We we have this channel of like user love and so whenever you guys share a success story we post it there and whenever you guys share like issues with our product we put it into our feedback channel. That way our broader team is able to act on it.
5090.0s
of like user love and so whenever you
5092.8s
guys share a success story we post it
5094.6s
there and whenever you guys share like
5097.0s
issues with our product we put it into
5099.1s
our feedback channel. That way our
5100.5s
broader team is able to act on it.
5102.2s
That is so cool to know. Thanks for sharing that. Well C, thank you so much for being here. Thanks for having me. Bye everyone. Thank you so much for listening. If you found this valuable, you can subscribe to the show on Apple Podcasts, Spotify, or your favorite podcast app. Also, please consider giving us a rating or
5104.0s
sharing that. Well C, thank you so much
5106.8s
for being here.
5107.9s
Thanks for having me.
5109.6s
Bye everyone.
5111.4s
Thank you so much for listening. If you
5113.1s
found this valuable, you can subscribe
5114.6s
to the show on Apple Podcasts, Spotify,
5117.1s
or your favorite podcast app. Also,
5119.7s
please consider giving us a rating or
5121.5s
leaving a review as that really helps other listeners find the podcast. You can find all past episodes or learn more about the show at lennispodcast.com. See you in the next episode.
5123.4s
other listeners find the podcast. You
5125.6s
can find all past episodes or learn more
5127.7s
about the show at lennispodcast.com.
5130.8s
See you in the next episode.
