How Anthropic ships the #1 LLM in the World safely: Separating the Brain from the Hands
Latent Space: The AI Engineer Podcast · 27 Oct 2026
Source: https://lspace.beehiiv.com/podcast/s/latent_space_the_ai_engineer_podcast/how_anthropic_ships_the_1_llm_in_the_world_safely_separating_the_brain_from_the_hands
Transcript: https://podcasts.opennash.com/shows/latent-space/how-anthropic-ships-the-1-llm-in-the-world-safely-separating-the-brain-from-the-ecd827/ (via OpenNash Podcasts, machine-generated)

Hi. Today, we're launch trying something new. We're launching coverage of Anthropic for the IPO. They just filed it as one as of today, and so we're releasing a series of podcasts leading up to the IPO, which is expected to be in two weeks for $2 1 trillion. And it's one of the fastest growing companies of all time, with a reported lifetime cash raise of hundreds of, uh, over $100 billion, and, uh, but a lifetime burn of less than $10 billion.

So, um, there are a lot of efficiency stories to dive into there, and I think that's something that we want to cover. How Anthropic is simultaneously shipping the strongest models in the world while also doing it very, very efficiently and building an incredible business. So, I’ve pulled out, you know, a a lot of, you know, if you’ve been listening to the podcast, that I have a former life as a financial analyst, and I pulled out some statements from the S one, which I thought would be interesting to give context to today’s podcast.

And we’re just trying this as a new thing because I feel like some of you want that personal relationship with us talking directly to camera rather than talking to our guests, which we’ve done for 4 years. So, the first thing I want to highlight is that Anthropic is an extraordinarily efficient company by human count, and that’s maybe obvious because it’s an AI company, so obviously they are using AI very well. Um, but I think the degree to which they have changed as a company over the past one year is not well understood.

So, the 1,000,000,000,012 month revenue per employee is now $6.6 million, uh, as compared to one year ago, it was $1.4 million. So, they’ve grown almost five times the, uh, revenue per em- per employee. And I think that’s something that we need to dive into more because I think any business owner, any engineer, you want to increase your revenue by per employee, not just to increase profitability, but also to increase the amount you can pay your people.

This is obviously tracked by significant increase in revenue per dollar of compute, and it’s notable that, basically, in Q2 of 2025, Anthropic was basically losing half a dollar for every dollar of revenue, and now it's making $2 for every $1 in rev- in cost. And I think that's an interesting, uh, anala- analogy of the, like, how enormously huge this inflection in Anthropic's revenues and business outlook has been over the past one year. Like, this year was extremely critical for Anthropic to execute well, and it has done that.

And I think we're gonna be basically gonna drive deeper into this and tell that story as much as we can from the podcast external point of view. Um, I The one consequence of this entire increase is that Anthropic has just reported its first profitable quarter, like, not adjusted profit, which is profit after we take out all the things that cost us money. No, this is generally accepted accounting principles gap profit, uh, of 19%, and that is enormous, again, completely flipping from one year ago, um, more than flipping, like, one year ago, it was losing 129%, uh, of every revenue, uh, dollar it gave it.

So, the primary attribution, I think, that explains this extraordinary efficiency, as you know, Claude loves loops, Boris loves loops, Dario loves loops, and there's 2 loops going on in Anthropic, the first is the improvement loop. Basically, there is the cycle of you have Claude, Claude leads to better models, more better models lead to more use cases, more use cases generate more feedback, and that leads to a better Claude. And then there's also the investment loop. So, revenue from, uh, expanding adoption leads to higher growth, higher growth leads to more investment, and more investment leads to more capabilities, that leads to a better Claude.

So, these two loops basically spin together and form an infinite loop that is driving the increase in efficiency until the end of time, according to the Anthropic as well. So, I think that is our transition from the overall efficiency story to how they execute it. And today, we're gonna focus a lot on the Claude platform because that is primarily, uh, one of the three things that they call the diffusion of Claude in the Anthropic as one.

So, they call this Claude Platform, uh, a comprehensive agentic development environment. They provide customers with direct access to the primitives to build, deploy, and scale their own applications that leverage Claude's capabilities within their own products and workflows. I think one thing that comes up a lot in the discussion in this podcast is the concept of app footing, where, uh, Claude is big a lot of Claude's products is actually built with the same Claude Platform that is available to you.

So, um, something that's also interesting in the Claude Platform that, uh, was disclosed, is over 50% of their customers with 100,000 more dollar or more revenue on a trailing twelve month basis use Claude Platform. And so, that is a increase and an adoption and emergence of Anthropic, not only as a model lab, but as also a hyperscaler and a Claude Platform, which I think maybe is in the name inherent in a Claude Platform, in a Claude Platform, but is going from theory to actual practice.

And this is something that we're focusing on in the podcast. The other thing that I also really am trying to emphasize, which I talked about before, is cloud code, cloud cowork, they all use the same primitives that Claude Platform, um, exposes. So, I think that's really interesting. Ultimately, it looks like Anthropic is really positioning their IPO to talk about going from chatbots to co-pilots to virtual collaborators. They use this term instead of agents, maybe it's more friendly, but it's a lot of the same things that you look for from agents, from long-running autonomous work to judgment, self-correction and iteration, and tool access.

So, owning a goal and working alongside of you instead of working inside of one app or task. So, that's ultimately the context I want to provide for this today's podcast. This was recorded when Opus 5.5 was launched, and I would say, overwhelmingly, among all the high taste testers that we talked to, Opus 5.5 is the best model in the world right now, and Anthropic is even just sitting on Fable 5.5 and probably will release it in the next two weeks.

I don't have they have insight information, but it makes a lot of sense for the IPO. So, let's look forward to that and learn a little bit more about how they build Opus 55.

Yeah, so maybe a distinction that I would make is in a lot of the incidents that people are hearing about and reading about, these are incidents that are happening to models, not just, uh, like everybody's models, right, that are in training. Where the way that you're running these models is you, like, put them in a box and you say, like, Solve impossible task, I'm gonna stress you out. Ready, set, go, right? And you, like, can only do what you can do within your environment, and the agent is basically incentivized to try to get out of this box and go solve the problem.

When people are running production workloads on our API or on anybody else's API, you're running an environment where, like, there are safeguards that are running that are, like, not necessarily running in training where, like, the whole point is to actually train the model, right? But in production workloads, first of all, you have those online safeguards, which are really great. Second of all, you have more of this architecture and infrastructure around what's happening, like, the agent isn't just in a box trying to get out, right?

Like, the harness is over here, the hands are over there, and, um, it's just, like, not even a problem that you would have to contemplate.

All right, we're here in the studio, the day after Opus 5.5, with Katelyn and Angela, welcome.

Thanks for having us.

Uh, congrats on launching the top model in the world right now. I would say it depends depends who you ask. Depends who you ask. Um, what is the main takeaway personally for you guys? And, you know, you have you have your official blog posts, the the pricing cards are are really cool. What do you take away when you launch a new model like this?

We're super proud of this one. Um, the team has put in a ton of different work to just go make this one happen. And I think we've, like, really heard the feedback on Opus 5's writing is, like, kinda convoluted. We really tried to clean that up, and I'm really glad to see that, like, it seems to be resonating with people. Internally, when we, like, tried out the model, we felt the same way, so it's just really great to get that validation.

To your point around the pricing cuts, there was, uh, a lot there, and I think increasingly people are looking for good intelligence per dollar. Um, and so making sure we get the per dollar part right, um, has been really important. There's some stuff in the edges that aren't really about, like, you know, maybe stuff that we haven't talked about in the blog post, which is that, we did do a lot of work on trying to make our API harder to distill from.

And so there has been some changes that we have made to go and make that a little bit more concrete, and these are things about, um, you know, like, how do we do, like, prefix-lock thinking, or how you can replay thinking tokens across different types of models. Um, and there's been a lot of, like, work and effort on our part around making sure that we try to adjust things for very valid use cases for where you might want to go and do this, like, potentially in your compaction strategies and things like that, versus in scenarios where, um, the use case of that is actually from, like, a bad actor, effectively.

So, there has been a lot of threading the needle around those fronts, and I'm glad to see a lot of that go out as well. Um, in addition to just, like, shipping a model that's, uh, good intelligence, uh, for for your dollar, and also, like, just communicating a little bit clearer.

Yeah, uh, specifically, you're talking about, uh, this paper, Stealing Reasoning Traces from Proprietary LLMs,

Yes, that paper. which is really as

That one. far as Right, they Yes.

obviously they did, like, responsible disclosure with you guys.

They did super responsible disclosure, and we're super thankful for for them coming to But, like, us and just like

could we, uh, for those who haven't heard about it, it's this is really cool that you could do it. I mean, it's patched, but, like, you know, like, what what was it like? Did it cause a freakout in in in Anthropic? Probably wasn't that big a deal, I don't know.

I don't know how much this paper specifically caused a freakout, because this is something that we knew was a gap. Yeah, so there Um, like,

there's 2 separate things. There's the Chinese model distillation, and then this was a specific technique that was published that was high profile.

Yeah, exactly. I we we've known for a long time that chain of thought extraction is one of the most powerful ways that you can distill models. And actually something like how do we preserve thinking blocks across turns and make that something that's more required is something that we've explored back for a while. And what we where we were at was we always have this really, really hard tradeoff that we have to make between user experience, like, user experience changes that are maybe painful for users of the API versus protection against distillation.

And so how do we kind of walk that line and say we're preventing distillation and maybe just to take a step back, the reason we don't want distillation is because the entire point of why Anthropic exists is cause we want AI to create positive outcomes in the world safely. And we run a lot of safeguards, classifiers, on the tokens that we're producing in order to make sure that they're safe. If someone distills our models, and they run those models without safeguards, classifiers, there could be bad outcomes in the world, and that's something that we don't want to have happen.

So, we don't want people to distill the model specifically for the sake of, um, just, like, using them for malicious reasons or for bad reasons, and so it's always an interesting tradeoff of, like, for all the people who are using the models for good, right? Like, we don't want to make them have a harder time getting the intelligence out of the models, but we have to make some tradeoffs in order to prevent that distillation. And so, we had been working for a while on how do we prevent that chain of thought extraction, like, what do we do about preserving thinking blocks in a way that wouldn't be too painful for people?

And we kind of think about user experience hits in, like, 2 layers. Like, one is just, like, you have to make an integration change, and it's annoying, but you could still accomplish all the same things that you otherwise would have been able to accomplish. It's just a little bit annoying. And then two is, like, actually, you can't accomplish the same things that you could have accomplished previously, and that's the one that we're, like, really, really, really trying to not have happen.

And so where we landed with the last couple models is a few incremental changes to have, like, enforce these preserved thinking blocks across turns. And now, like, starting to be more about, like, across, you know, like, if you're using multiple accounts and things like this, right? Like, how do we make sure that we're, um, ensuring that you aren't able to take advantage of this chain of thought e- extraction with the models? And so we shipped a couple additional features to help with that.

We worked actually with a lot of open source harness maintainers. When we reached out to them, we talked to them, we were like, You're gonna have to make these upgrades within your harnesses. We wanna make sure you have a good time. As well as, obviously, our big customers, like, who are building harnesses on top of the API. And for some of them, a really good example is they have compaction strategies that would have been broken by the need to preserve these thinking blocks.

And so we shipped a new, um, compaction feature within the API to replace what they had otherwise been doing manually, but still works with preserved thinking. So, this is an example of, like, we're walking this line, but we didn't just, like, kinda YOLO out, like, now you have to preserve these thinking blocks in this exact way, because we would have broken a lot of people, but we were able to take the time and work with customers and open source harness maintainers and give them what they needed to be able to make that upgrade.

While we're on the topic of it, anything on the other side of distillation, so, uh, you know, the other half.

Yeah, no, we we have seen that it is harder to distill, like, through the measures that we are able to to take a look at, it has been significantly harder, cause this is, uh, and you know, credit to the the researchers here, they they did disclose it, uh, very responsibly. Um, and so closing it down was was great, um, and they did publish their results, uh, very publicly, and I know it it did kinda trend quite a bit on Twitter, um, so many people could could use it, uh, but, yes, we have seen that that actually from a the areas that we are able to measure, um, it does seem to have, uh, reduced the effect, um, the negative effects

that we don't want. And, yeah, the bit I would just add to what Katelyn just said is, I think, uh, at least I see a lot on Twitter or X or whatever you wanna call it, um, it's a commercialization problem or something like that, you know, if if you can steal a model and you can take this, then it's you can run it for cheap and and so on and so forth. While true, um, it's actually not something we really care about very much.

We care much much more about this from, like, a safety point of view, um, more than anything else. And so that's why we we tend to think about it from that angle more than just, like, commercially. Realistically, if you distill from us, we you actually pay us. So commercially, we, like, make the money. Um, so it's not really that kind of an impact. Um, it's just more on, like, making sure that our models, like, we wanna make clo- we wanna make sure Claude is, like, good, and Claude could be derived, and if it sends a negative, that would be unfortunate.

Just generally, al- also on F5, cause you you guys are just fresh off this. One thing that's really fun to establish the dan- dynamic between you 2, because you also do a lot of appearances together, is you are running, uh, you're head of products, you're head of engineering. How does this play between the the 2 of you, uh, especially for a model launch? So I'll just take an example.

So for model launches, so our team works together very closely, um, like, between product and and eng. On my side, a lot more of it is about, like, what is the balance of trade that we need to make between, like, launching something that has the right, like, intelligence bar, but also the c- the cost to preservation, um, like, a lot of, like, our our updated cache costs and things like that, are indications of us being, like, we want to, like, encourage you to take integration changes that are beneficial for actually everybody and just, like, pass the savings on.

Um, but a lot of these kinds of decisions around that, and then how do we balance, uh, some pieces around our safety motion, as well as making sure that it's, like, an awesome model that people can can take advantage of. And so those kinds of trades around, like, I would say maybe the qualities and surrounding components around how we launch the model are more on product. And then from an integration experience standpoint, a lot of that is from, like, the engineering team directly.

So we work obviously, if you're, like, talking about the experience of an API that's, like, very, very, very tightly integrated, and so it's actually been really awesome. We have a lot of engineers who actually are deeply product-minded, and we actually, like, love that. And similarly, on my side, um, there's a lot of, like, product managers who are much more technical. And we want this, like, blended kind of interdisciplinary and type of people, because then they can kind of analyze it from all the different angles to try to figure out how how, uh, to do that best.

But I tend to lean a little bit more on, like, commercial impact versus, I think, I would actually frame it as more of, like, the engineers are going to care much more about the integration experience. Um, so they almost take more of a UX angle to it, um, where the UX is obviously, uh, more about the API, um, and the developer tools.

Yeah. It's fun to be a product engineer who works on developer products Yeah, the developer and and platforms.

platform, yeah.

So, I would echo that. I think we have a lot of engineers who love thinking about the product, who love thinking about the user experience, who love thinking about the developer experience. I think it's a dynamic that we have between ourselves too where, um, I have a lot of opinions and am annoying with our product team on how we think about these things, and our product team is also really technical and they can get in the weeds of the decisions that we're making for scale or for whatever it might be, um, which is a fun dynamic that our team has.

Yeah. Something, uh, Katelyn and I have worked together for a long time, um, and Stripe and I at Stripe, and then, uh, I guess we did not work together at OpenAI. Um, Do separate but uh, just

that's just you. labs.

separate labs. Um, and then I came over. But, uh, I think some one interesting property of both of us has been Katelyn frequently actually gets mistaken for a product manager, and I frequently get mistaken for, like, someone in engineering. Um, even though, like, those are not actually our job functions. Um, and so I think that pairing tends to, like, help out a lot with just making decisions that are consistent across both sides.

So I I think I come across a lot of companies with a similar, like, product leader, engineering leader setup. As software engineering gets solved, maybe those 2 roles merge.

I kinda think so. Yeah. I think everyone tends But if to lean towards

you guys are, you know, separate still, then we we're not there yet, because I

We're definitely not there yet. Um

so so, like, why, right? Like, what's left?

I think that, like, software engineering getting solved is I would I don't know if I would ever use language quite that strong. I think there's the way that I tend to think about these things is, like, ambitions will continue to elevate, right? And maybe a good example, like, a lot of what our team is spending time on right now is, of course, obviously, all the things that we have to do to launch our model safely is actually the most important thing that we're doing because of all the work that we have to do on security with the agents, with offline safeguards and online safeguards, all these things that are happening.

But the other half of what we're doing is trying to scale. Um, and part of that is because when I joined Anthropic a little over a year ago, we had the Messages API. It was a Python monolith. We were running it in a, you know, single GCP region, and everything was fine. And none of my friends and family, like, knew what Claude was when I was going to take this job, right? And now we're in an extremely different place.

And so the metrics are really wild in terms of how much volume we've seen and have had to support on the API. CPU is really constrained, uh, which has been a really interesting one for us. Um, and so we've done a lot of work on how do we scale these systems, and that takes a lot of, obviously, just, like, distributed systems expertise and certain language and frameworks expertise that the models have. But when it comes to operating and evolving a system at scale like this, we're not quite we're not nearly in a place where it's like, Good luck, Claude, have fun, uh, and just go kind of make all the changes that you need to make to the system, roll

them out safely and operate this thing at scale. Um, there's still just a lot of work for humans to have to do that I imagine will continue for a while.

I'm curious how that's changed. So, like, a while ago, you know, when Fable first came out, we had Glasswing, and there was a whole delay and change in how models started to get released. We saw this for most frontier models. Uh, there's safety testing. What's the status of things now? How do you guys see it? How do you determine, like, one thing is there's a new Sonnet and, um, Haiku also coming. I assume those will just come out.

But for something like Opus 5.5 or a new Fable, what's the what's the current status?

Yeah, so I would say this starts with what our kind of policy decisions need to be on safeguards and on safety. Um, and that's both offline safety and online safety. And so there's been a lot talked about lately around data retention and what we're doing there in order to, what, you know, release these models safely. And the reason for that is because you have, you know, situations where there are many, many accounts across, like, a whole giant network of people who are choosing to spin up accounts and use the models in various ways.

And there's no way that in an online way, as tokens are flowing through the system, we would be able to kind of catch that there's something malicious happening with the models or distillation or whatever might be happening. And so we've started to have to have data retention policies so that we could do automated review and human review, whatever it needs to be, to make sure that good things are happening with the models. And with each model that we're releasing, like, that's one example of a policy decision decision that needs to happen on, like, what will the policy be for this specific model.

Um, and so Opus 5.5, that launched yesterday, for example, does not have those data retention requirements that Fable has, right? But that's a decision that we have to make each time we're launching one of these models. And if it is going to have those policies, we're gonna have to do the work to make sure that those things are enforced. Similarly, with programs that we have, like, we have cyber verification programs, and now newly our life science verification program, which says, you know, like, you have a difference in how the online safeguards are working for the model that you're using, making sure all those programs are in place is another one.

And so, it's just kind of a flavor of the type of work that we've done as part of each model launch, that these are everything that I'm saying right now are not things that we used to have to think about when we were launching these models. Um, and now it's it's the majority of the work is, like, what are we doing to prevent distillation? The online safeguards, the offline safeguards, the policies. A really great example is we built a system to make sure that authorization for being allowed to use a model or being allowed to use one of these, like, programs that give you maybe, like, relaxed safeguards, that authorization threads all the way down through our system in a

way that we can really trust. You know, it's not like, Oh, you got access to Mythos, and so we throw you an API key and we flip a feature flag and call it a day. We have a really robust system on how we're doing auth and making sure that if you're supposed to be using that model, like, only you are using that model in the way that you're meant to be using it. So, there's a lot of complex things that are starting to happen around model launches that, again, we just never used to have to think about.

Oh, that's good.

Yeah, and it's it's scaling up with capability. Like, as our capabilities get w- uh, more and more advanced, and they start to spike, you know, in specific areas, that's where we need to pay, like, extra attention. And so a lot of that evaluation comes just from, like, those very particular cap capabilities that we might either see as, like, offense-dominant or defense-dominant, and therefore might take different strategies for all of that. And as we bake that all the way through to eventually the model launch, there will be components kind of like Caitlyn mentioned where we kind of don't exactly know how people might use those capabilities.

There are some more obvious ones, like, between coding, right? Like, great, go make awesome products, go ship more things. Um, it can be really awesome for you. And then other areas, like bio and cyber. It's like it could be dual use, and there could be negative effects, but obviously, of course, you could be a really valid cybersecurity researcher. You are trying to find vulnerabilities, you are trying to be a good actor. Similarly, with bio, you might be researching and stuff like that.

And so a lot of these programs, like the one that you you have up here, are to basically provide access to that dual use. You do have to specify your use case and things like that, um, but we do want to enable the community, uh, to be able to take advantage of these capabilities and see what amazing things that they can do with it as well.

Well, we'll put up the links, and I maybe we'll dive into them later. There's also one for bio. Also, I just think at a high level, you know, it's it's really important that you guys actually think about this stuff. People forget, like, Fable was rolled back, and there's all this distillation,

It is.

and people are like, Ah, zero data retention is not a thing on Fable, crazy. But you have to remember, like, you know, it got pulled back. Uh, there's distillation attacks. Like, it's good you're overthinking and doing it in the right way while releasing it, right?

Yeah, and a lot of that is, like, building the right framework, um, just to again, kinda talking about the edge side a bit, building the right framework in our system to be able to do that. So I talked a little bit about this, like, this idea of, like, authorization and trust grants that we've built throughout our system now.

Never say never, right? Not not as not as precedent, uh, but also it's, like, a very common thing in platform engineering that you, like, uh, you have one incident, and then you, like, overprepare for that incident, and then incident never happens again. Which is, like, it's like, but you

It's true. Which is why,

never know.

you know, now you have the button.

Well, it's true, and it's funny you say that, because it was a like, with everything going on, we then had this line item in our, you know, the work that our team was doing that was, We're gonna put this system in place that makes it so that it could be a single button to pull back for everybody. But, um, and so it's funny cause it's like, do we justify that we're doing this work? Cause we don't we hope that something like that never happens again.

But I think in this case, it's really just a laddering up of those same primitives that we wanted to put in place everywhere to help us manage these verification programs and access to models like Mythos, and otherwise to just say, like, just set that same thing below, like, everybody's model access, so that in case we ever needed to, we had the right controls in place.

Yeah, and decisions like this have actually proven to be, like, we'll see about this one in particular, but examples like this have been consistently the way that we've had to actually transform the way we develop our roadmap. Because, you know, in traditional product management, I guess, you can be like, Okay, like, these are the set of things I hypothesize to happen. They will have, like, a s certain sequence, a certain ordering, but you are picking bets in a way that are much more concrete versus, uh, like, you know, the de-deployment of Fable, for example.

I don't think any one of us would have guessed that that would have been, like, a valid use case that would have happened, um, by the government, and the way they just, like, had to go do it. But, clearly, the probability is non-zero. And so from this form of thinking, it's become deeply probabilistic, actually, um, from, like, a product management standpoint of, like, So should we prioritize this? What are the chances of it ever happening again?

When we look at things like that, we're like, Okay, if we are to imagine the most crazy set of possibilities, we probably actually have very little s certainty as to where things will go. And so we actually just kind of need to, like, hedge and build a portfolio of bets. And so more often than not, we will take hedging scenarios of, like, Okay, we don't think it would happen again, but in the event that it does, we would actually want that hedge or, uh, have the ability to be able to do that very well.

And so we will prioritize things like that. And then we have increasingly found that more often than not, by taking that strategy of product development, most of the time, we actually find ourselves being like, we actually do use the thing.

Is there an, uh, another example that's a separate one, just so I can understand?

For example, this this access program. This access program is a really good one. This, uh, it's internally, it's it has a bunch of granting and all these kinds of things. We developed it specifically for reasons about, like, being able to actually give programs like this out to customers, like, very, Do you wanna very specifically.

give a high level what the access program specifically is?

Sure. So, if you want dual use, if you want so for let's let's take a model. So you take a model, uh, call it, like, Opus, um, and you want to be able to use advanced cyber capabilities on on Opus for research or for whatever you wanna go do, like, you're you're a defensive user. Um, you can apply to this program to be able to with your use case, and if approved, what we'll do is actually diminish the safeguards on your version of the model so that you can then go ahead and try to take advantage of th- those, like, cyber activities, um, for for your use case.

Now, obviously, that requires, like, per-use case understanding. It requires scoping down to a particular user, cause we wouldn't do that generally, you'd have to, like, apply. And even things like this, like this kind of program, we designed a lot of the primitives that Katelyn was talking about to be able to handle things like this. Now, it was surprising to us or maybe, like, i you know, in the portfolio approach, we found that we were actually able to take this and make it ex- we were like, when we're designing it, we're like, What kind of should it be?

How generic should it be? Should it not be too generic? Should it be overly focused on making this verification, you know, style portal plus granting of a very specific capabilities of the model? And we tended to opt for, like, more genericism because we were like, We don't know. Like, maybe we could use this for other things, maybe not. And then it ended up being massively useful for things like, um, you know, uh, access to the model writ large, for managing, uh, any of our trusted access programs for handling de-deployments of models, because we were actually very, very generic about how we handled every single primitive.

From a normal product standpoint, I would probably normally probably be like, That probably doesn't make a ton of sense. We should focus on the use cases that make sense. But here, increasingly, because we choose the more generic approach, we find ourselves reaching for that solution more often than not, and it pays, like, increasing dividends for that reason.

It's, uh, something that you needed to have anyway, and you just have to force your function to have it. On a technical basis, what, uh so I am f- I'm from the AWS world, right? I I am an SRE But and then people, like, uh, who get really intense about this, talk about SANs, a bar, and all these all these other technical architectures. I'm not familiar with the word token grants or or, like, uh, uh, trust grants, so trust grants, trust grants.

Is there a difference between normal off see and and what trust grants are?

I mean, not really. It's it is it's a it's a form of authorization.

Yeah, it's I am, but and Anthropic Yeah, kind signed it.

of. I think so, really, we have, like, um, there's, like, authorization of the form of, like, okay, you're you made a request to an API that, like, under the hood, is gonna access some data, and we're making sure it's, like, your data that you're accessing, right? Like, really, really vanilla authorization. This is just, like, we put a data model in place that's, like, a very clean, like, a trust grant means, like, you have been given access to some specific model, or you've been given access to some specific program.

Or, you know, like, there's a a bunch of different permutations of this. We actually find ourselves reaching for it all of the time. Um, another good one was when we first rolled out, um, some of this preserved thinking in, uh, or for the for Fable 5.1, we kind of grandfathered in some people and gave a bit of a, you know, just, like, a

Grace period.

time yeah, a grace period to go and, uh, make the integration changes. And as part of grandfathering in, we said, okay, we have to, like, track who is, like, an existing customer who needs the extra time, right? And then we did a trust grant, and so then we're able to say, for anything, we're just running this service that is kind of on the hot path of our requests to make sure that, um, you know, we just, like, flow through and and check whether you have the grant for the thing that you're trying to do.

Um, which is it's really not any different than, like, vanilla authorization, really, but, like, we just had to spin up a system that was specific to our system and, like, where we needed to be able to insert these checks. Yeah. I mean,

uh, I think, uh, these are all, like, the engineering things that people kinda take for granted, don't don't really think about. What is another example, I guess, of the platform stuff that had to happen, you know, maybe serving Cloud Code, uh, maybe serving, you know, obviously, I talk a lot with Cognition, but other customers, just in terms of a model launch, right? Like, we'll talk more about the, uh, the the the safety work that is that's being happened.

I just wanna sort of go at a high level, because it's fresh off your memory, like, what goes into a model launch?

Yeah. It's become a bit of a a machine these days, uh, minus the safety stuff, which I think is is the more complicated part. But outside of those pieces, there is a lot of getting feedback from our customers. So, you know, we make the model, there's a lot of safety testing, um, there's also evaluation of where we think it needs to be. We make the required trade-offs and decisions as necessary. And then we have to go through testing.

Uh, we'll cut various versions of the model, we'll evaluate it. We'll do a lot of internal, um, ant feeding, uh, is what is what I call it. And that actually always really helps. Like, we catch, like, really early things like, Oh, this is, you know, maybe not the best. And we obviously use Claude a lot for our work internally, so we're able to actually see early signals around stuff, um, and helps us detect, like, you know, which which variants might be better than others.

And then when we feel really confident with, like, a certain set of of variants, uh, we will cut a version for our for our customers. Um, sometimes that will go through our first party products, uh, like through Claude Code or Cowork or things like that. Um, in other cases, it will go through our API customers, um, and we'll engage in a kind of, like, a, uh, early access program, effectively, with those customers. They'll need to make changes into their harness, and we've engaged with, uh, a lot of customers, like, like Cognition, like, like Cursor, like these folks, and we have a bit of a process on how we handle that.

Um, so that we all kind of get they all we get the feedback that we we need. Uh, does it perform well in your evals? Do you like it? What don't you like? Some things are are vibey, and we like to hear those things too. Um, and some things are just like straight-up evals, like, Did it do well on the things that you care about? And we take that all together, and once we f- all feel good about, like, all those different pieces, we launch the model.

The state of model feedback is feels very unstructured still. Basically,

It's clear.

it's always a Slack channel between companies of, like, Hey, this thing seems weird, and then people are like, Okay, I'll fix it, I guess. Give me an ID or something.

Yeah. It's very I mean, it's it's the nature of, like, a product,

Can we do better?

it's, like, non-deterministic, right? Like, but I think to your point, the only customers really that are part of our early access programs are customers that have really robust evals. Like, I do think a lot of this is vibes-based, and I think a lot of the conversations tend to be kind of, like, vibes-based and and on the fly. And part of that is because we're, like, racing, right? It's like we wanna minimize as much as possible the window between we have a version of the model that we've cut for testing and we give everyone access to that model.

And so it is just like this really tight window. But the customers we tend to do with are the ones who, like, we know they're gonna go, like, kick off their whole eval suite and be able to come back to us with, like, really actionable feedback as quickly as possible, alongside the vibes um, the vibes-based feedback. So, um, of course, there's, like, definitely, like, process-y things that we could do to make these things better over time.

But, I mean, you'd be surprised. I mean, you probably wouldn't be surprised, but I think some people would be surprised about how few, you know, like, products that exist in the world that are agentic that, like, don't have good evals to Oh, we be able to help test these had we had models.

Boris on the pod two years ago saying we don't have evals for Cloud Code. Uh, I mean, that's changed, but That has the that has changed. changed. But, you know, like, uh, yeah, a lot of things, like, you know, your evals are your users, and you don't have time to formalize them, or you don't have the the experience or whatever, and that that's fine. At some point, when you get serious, you'll you'll have evals.

And I think the good thing is you guys do a lot of app footing, as you say, so it's commonly said that you find a lot of this testing it yourself before it comes out, right? Yeah.

Yeah. The API s um, customers in particular are very useful because they'll put Claude into a different harness. Um, and we want Claude to be very, very general. Uh, so its ability to work in other harnesses is is quite important to us. And so that style of feedback tends to be exceptionally useful for us because it's it's novel, um, from what we would have gotten from ant feeding.

What kind of feedback helps? Is it, like, regressions? We have regressions here. We notice good attributes here. Uh, the good Yeah.

stuff is always it's always very validating. Uh, I think we're very frequently not surprised, but we like to look to be validated on some of those things that we might hypothesize. Honestly, the most useful feedback is negative. Um, like, you know, it didn't do well at this, and do well at that. We love that feedback, and then we're like, Great, how do we fix it?

Yeah, and it tends to be some of these things that are, like it's really hard because not everyone actually wants, like, Claude to land at the exact same spot on every spectrum. And and maybe a good example is, like, you know, laziness versus proactivity in taking action. And so there's sometimes when, like, we have a model, and and people are like, Ugh, like, just finish my task. Like, stop being lazy, like, do better. And other times when it's like, Hey, we were, like, just kinda feeling this out in planning.

And then you, like, went and wrote a bunch of code, or you went and, like, took a whole bunch of action that I wasn't ready for. And getting that exactly right is literally impossible because everybody actually wants something different in reality. But I do think we've had situations where we've cut an early version of the model. We found that we're we saw too much laziness, right? And, like, we, like, put it back in the oven for a bit, right?

And and tried to train that back out. And so those are the sorts of things that we may not have caught if all we're doing is internally ant feeding within our products and within our API. It's way, way more important for us to get that out there in the API and get our all of our other customers running with it and let us know what we need to change.

Yeah. This is a core question. I think it's probably the i if I if I were in your shoes, like, one of the hardest things about PM-ing for a model, which is, uh, let's call it alignment versus personalization. There's a default personality, a default behavior that you want for all consumers. And then you want to be able to respond very quickly to someone's Claude MD saying, No, no, no, like, please be more proactive than default. Uh, and then, like, so as to my knowledge, you know, and I was involved in Frontier Code, uh, we were involved in the Fable launch, uh, which put Frontier Code on the map.

Uh, and now Opus 5.5, we have it back on on on the the the model benchmark. That is a very big part of Frontier Code, right? The the reason that the model curve looks weird for Frontier Code is because it pen- penalizes too much activity or too much, uh, files. Uh, which is good. Like, that was the whole design purpose of Frontier Code. But, like, I could easily just be like, in my Claude MD, just say, like, don't do that, and then it won't do it.

And so, like, am am I penalizing Cloud Opus 5.5 unfairly because my benchmark doesn't have that personalization element, which is subjective. So I don't know what the question is, but, like, how do you deal with, like, per personalization, right? Like, like, is it, like, clearly a model can do so many different things, and it's so depend everyone has their own Claude MD. Uh, how do you eval that?

We don't do a great job of it. There's, like, the there's, uh, seriously, it's it's really hard. It's super hard, and they're they're competing

Yeah, look look at that.

and contrasting.

It's it's horrible.

I hear I thought I thought there's the Claude Constitution, I think That's old. But there there is I I use a constitution, but I'll I'll let you guys answer, you know.

the Constitution is helpful, but but for things like this, on how it personalizes to instructions, uh, which are different and preferences, which are very, very different across these things, um, is very hard for us to do. We generally try to solve this by having more parameters that then allow us to tweak it per customer base. And if we don't control your customer base and you're an API customer, then we try to give you recommendations and then you try to do that tweaking.

So there are things like, you know, around doesn't exact it's not exactly the same level of like personalization to the Claude.on, you know, various Claude.on MDs or now agent MDs or whatever. But we do try to give things around the layer of, like, level of effort, like how much effort you should put in, like, what are, uh, the, like, the level, like maybe the size of the model that you might actually, like, try to care about, things like that.

But it's I think it's, um, it's not quite there for what we would actually need for, like, that level of personalization. Um, so it's a very hard problem. And right now, a lot of our way to solve that is beyond the parameter, like the giving multiple parameters to something, is try to exhibit conflicting taste, honestly, across, like, all of the feedback that we get, and try to try to make, like, a judgment call on some of these things.

I think over time they actually and as you scale to more customers and more users, that's, like, not a very scalable way to go do something. Um, and we will need to develop ways for Claude to kind of, like, take in your personalization. Um, and maybe that's, like, a signal you can provide or something you can tune on or things like that. Um, but I think we'll have to, like, take a route that's much more oriented, uh, towards that direction.

Uh, but for now, yeah, no, we we actually just try to make judgment calls on on a lot of this stuff, and it's it's pretty tough. It is what it A small is. Uh

a small note there, I guess the default effort reasoning changed. It's now medium. This is one of those things that probably affects the average customer that doesn't use this toggle.

yeah, I say like a cloud code setting, right?

It's a cloud code setting that defaulted to medium. They just Yeah. use And and this and this thing varies. Thanks. This thing does, like, this is an this is a perfect example of a parameter that we use to kind of, like, personalize. I I would say this may be a little too aggressive of a term, but we kind of, like, customize, maybe, for different subsegments. So, like, uh, the default setting will adjust for, uh, whether you're a free customer or a prosumer customer, Claude Code versus, uh, Cowork versus enterprise, like things like that, versus API and stuff like that.

Um, knowing that you can obviously always adjust it, but those are kind of, like, light ways where we try to give that. One thing that we are hesitant to do is we don't want people to feel like they can't control their customization of Claude. Uh, so we could take a heavier hand on these things, for example, but we want to give users the ability to experience the intelligence levels of Claude as they kind of see fit.

Especially if you wanna go towards, like, the higher end or you wanna experiment with, maybe sometimes in the in the lower end, if that actually is, like, faster for your use case, which is a hard product tradeoff, I think. Um, and so we we tend to lean towards gentle defaults, but give you options to to adjust, which I think works okayish for now, but I imagine us, you know, like, learning how to do that better over time for customers.

We're not, like, super, super happy with any of this yet, but we know as we launch a model, like, you you ask, like, what does it take to launch a model? One of the big things, obviously, that we have to do is, like, okay, Claude, like, got good at a thing. And, like, sometimes that thing is something we should expose as a parameter on the API to let you control. But I think as this has evolved over time, we've ended up with, like, this funny mishmash of, like, there's effort, there's budgets, there's all these things that you can set on the API.

And it's like, how could anyone, like, possibly figure out what is the right combination of things for any given task for what you should go and set? And it's something that this is us just very much, like, figuring it out as we go and figuring it out with customers and figuring out, like, which of those parameters do we deprecate over time, which of those parameters do we just, like, default better, which of those parameters should be, like, a 1 to a 100 scale versus a low, medium, high scale, right?

Like, because actually, we've trained Claude to get really good at understanding, like, a whole spectrum of options on some of those sorts of things, but, like, do we expose the, like, entirety of those millions and billions of permutations of options? Like, you wouldn't really get very far. And that's one of the things actually that, um, is part of the value prop with something like Claude Manage Agent, which is, like, still programmatic access to Claude and an API, but you get a harness out of the box where we do every time we're launching the model, a lot of the work to go figure out, like, what are the new parameters, what are the breaking changes, how do we get the best

out of Claude within our harness. So that for the people who are using it, um, you just kind of get that out of the box. You don't have to think about it.

Just for people who are, uh, dr- drawn to that, at which there is a significant and growing population. Specifically, we actually did a whole CMA workshop at at EIE, which I thought, uh, was really well done. Is there a ch- separate feed for people to subscribe to, like updates on CMA? Feed, like, um Like, uh, Twitter feed, newsletter, I I don't know. No. I Like, Chase think actually our blog.

our Claude dev channel, we

That's it. Okay.

are yeah, we're we're pretty loud about it, um, because it is, Yes, it's it's like, an interesting it's Claude's, it's option for people.

Claude's. Okay. Yeah. I just wanna give people, like, if that's your if that's your way of consuming Claude, that's the this is the path. I think Um

it's also pretty underrated, um, for those that aren't consuming Claude this way, what is Claude managed agents? Why should people use it?

oh, yeah, we should probably do a do a little intro. I don't know, they've done so many talks. Yeah. I But maybe don't know if a little we've talked about bit, a little it, but bit too too sentence thing.

I think a lot of people glanced over it when it came out, right? Everyone was in this phase of I need to do harness engineering, I need to control this. And, like, you hear all that, and now it's like, do you need to do all that? Or do you just use managed agents, and you have a lot of infra you can skip over setting up? But give give the quick

yeah, it it depends on your use case. But, um, so Claude managed agents is our API today. We have the Messages API, which is essentially raw token access. Then we have Claude managed agents, which is a suite of APIs, but it lets you define an agent, start a session with that agent. Um, and it's essentially just kind of a higher order abstraction that lets you skip setting up all the infrastructure, setting up your harness and doing a lot of that type of work.

It's a much more out-of-the-box integration for people who want to programmatically run agents.

Does CMA, uh, share infrastructure with Cloud Tag? Because the sequencing of it seemed very, like, oh, like, that's the Claude 3.5 Sonnet, is a lot of it,

um, a lot of it. So, one of the big things that we do is as the platform team, we run our APIs. We also kind of run all of our foundational product infrastructure and product services. So, like, really good example is, like, our team runs our sandboxing fleet that backs both Claude managed agents as well as products like Claude Tag and otherwise. But at the actual harness layer for those two products, like, there's just differences in what we might wanna do within the harness.

So that they don't they don't actually share a harness, um, but they share a lot of the infrastructure that powers the execution and, um, powers the APIs.

I think that's that's really, like, people should use this more. Uh, I think everyone basically had copycat clones of managed agents after you guys launched it, which, uh, you know, imitation is a highest form of flattery. Since we're on this topic, we'll briefly mention, uh, all the memory stuff that, uh, that has been going on. I feel like it was a big part of Anthropic's work in the first half of the year. Has it stalled out, you know?

Is is it still still ongoing? Like, what's what's the emphasis here now?

Manage agents specifically.

Memory, dreaming.

dream, yeah.

Oh, it is a big part of what we're doing. I think maybe the way to think about it is we launched managed agents as a platform for people who, like, didn't want to do all these things themselves, right? Or just wanted to, like, offload, like, I'm getting the best out of Claude because our team is literally sitting here, like evaling all these things and making sure that the harness is gonna be really powerful. Memory and dreaming is all about just, like, how do you make sure your agent can continuously improve over time?

And all of that is still an important focus for our team. But what I would actually say is, to the point of what we talked about earlier, as launching models has become harder and harder from a safety perspective, I think most of our team's energy is going less into, like, Here's a fun feature to, like, hill climb on this agent or whatever, and much more into how do we bring safety into all these things. Because the most important thing that we can do is get the power of the model out there and get it out there safely.

And so I would say a lot of our energy is take all these same concepts and give you an agent system that's gonna be really powerful, but also that's gonna be really safe. And so that's where we're putting our energy.

I think compared to maybe earlier this year, we have slowed down on, like, skills, dreaming, memory, like these kind of, like, moment-in-time

Please don't kill the skills. objections. No, no, no, It is we love skills,

on self.

we love skills. Um, we we are making we are making updates. Uh, I'm not saying that we're not doing anything, but, um, we we love skills as well. But, yes, I think we've slowed down a little bit comparatively, um, for some of the other efforts. But, uh, we are very bullish on, uh, people kind of doing some level of hill climbing at the harness layer, uh, to, you know, some of the previous conversation here. We do a lot of this ourselves, but the harness that we provide is very simple in a way.

It's it's intentionally simple. Uh, we don't want to be actually overly opinionated. We wanna make sure the best practices get merged in, and there's some kind of undifferentiated pieces of the harness, uh, agentic loop that, yeah, are, uh, hard to string together, and then also, like, you can kinda just boilerplate do it. Now, the interesting part on top of that, though, is, like, once you can get that base layer, there are these, like, hill climbing strategies that we construct.

And, like, dreaming is actually one of them. So dreaming under the hood is a multi-agent system that we hill climb on top of CMA, and it's actually age CMA on top of itself, basically.

CMA on top of itself.

It's it, like, spins up another yeah, uh, the way Dreaming works is that we spin up another managed agent that then can inspect the previous managed agent, and then write, uh, memories to that managed agent. So, it's managed agents all the way down. And this is a bit of thing by design, because we envisioned that you could probably put together a lot of these modular components for something interesting. And Dreaming is just, like, an example, one of the more interesting things that we were going after.

Uh, but we kind of have been looking at this and seeing, like, there's actually a ton of other stuff you can do. You can do, like, best of n as a type of, like, multi-agent strategy. You can do a verifier agent, uh, that kind of, like, goal thing or outcome thing. Um, there's a bunch of other stuff in there, and we've seen people do some pretty funky, crazy stuff. Like, there's been a harness that, like, shares, like, crazy memory across everything, and they spin up, like, tons of parallel agents.

And we're like, we should actually just probably enable you to do that, uh, in the same way that we did we built Dreaming, that's, like, managed agents all the way down. There's a update that we'll make relatively soon, um, where we're updating the underlying primitive, we'll expose it, uh, to allow users to kind of, like, construct harness strategies, maybe, or however you wanna call this, Like, multi-agent like, multi-

strategies, yeah.

agent strategies that allow you to kind of build your own version of this, and we'll give you, like, you know, templates of, like, this is how we started with Dreaming. Maybe there's, like, an update to this that you would wanna make as a developer, and you would find some more value on top of Yeah, I it.

think that's a really good framing. It's less that we're slowing down on this sort of stuff and more saying, like, we've got the primitives in place. We can continue to let people just, like, take and run with those things while giving them the tools to say, Here's some, like, cool multi-agent strategies and, like, things that we've put into the into Claude managed agents to make it really powerful. Now you can, like, let Claude do work to take advantage of those primitives and make it even more powerful yourself, like, create new multi-agent strategies and whatever it might be.

While a lot of our team's attention turns to, Okay, great, like, community, run with that, right? And and you do you while we continue to focus on, what are all the things that we need to do to make sure that you get the access to the models within that system safely.

One thing you brought up was you run a lot of evals on this stuff. Pretty hot topic right now. Any advice, best practices?

For eval, specifically?

Evals on harnesses, agents.

There's a lot of different types of evals, right? Like, not all evals are made equal for different, you know, like situations that you're in and things that you're trying to test for. You obviously have just, like, static, deterministic stuff, and then you have LLM as a judge. You have all these different strategies. But I guess I would just say, like, what we found from most people is, like, you probably need more rather than less. Um, it's just, uh, there's not, like, one perfect, like, couple sets of evals that you could put together that are gonna get you all the answers that you need to make your system better.

We found that people end up needing, like, a relatively large suite of these things. And then it's like you can have your agents work with you to evolve your eval suite over time as well. Um, but I think it's just, like, n- under investing is maybe, like, the biggest mistake that we've seen from people on that front.

Yeah, and let's start simple. I think some people, like, go a little crazy and they're like, you know, they're like, look at all the the menu of options that you can kind of listen and you're like, Wow, this is so crazy. Let me try this and that, and what's the best one? And, like, while they do all those things, like, the number of evals they actually have are, like, zero. So, you might as well just, like, do something simple.

It will help. Um,

Yeah, I think in some some part of people are are inexperienced, they're not trained to do this, um, and they don't know what good looks like. So I think for you guys, publishing more examples of that is is really helpful. Uh, something I I I do tell people a lot, uh, that I'm doing with more fable and astro class type models is we're treating engineering like research. And when you treat your engineering like research, you naturally do more evals, you do more labeling, you do more scaling of runs, or, like, prototyping and scaling of runs.

And you're not training a model, you are building the harness, the product, whatever it is, but it you're approaching it like research. You're defining upfront what your, uh, goal is, trying to formalize it as an eval, and then, uh, trying to identify, Okay, why why is it failing? And sometimes your eval is wrong. Actually, a lot of the times, your eval is wrong. Sometimes your environment's wrong. And then sometimes the model is wrong. Uh, and and, uh, actually, that is, like, a lot of part what I think AI engineering is becoming, which is doing all this stuff.

Completely agree. Um, and

we'll continue to launch more features within Claude managed agents to give people more tooling to do this sort of stuff out of the box. And you'll probably find from us that you'll get something that's like, Here's a product that will help you create evals for your managed agents or whatever it might be. And it'll feel very much like working with Claude in order to do that. Cause, like, that is that is the best way to do it.

Claude's actually really good at creating evals to, um, And eval evaluate agents. UIs, right, actually, Yeah, exactly.

because, like, people, like, blur blank out when they look at, like, JSON Ls, like, just right, but a pretty eval UI where you're, like, left, right, left, right, left, right, then you're you're sort of, like, doing what the human greater data people labelers would do anyway. Uh, you can do that yourself.

You can. You just, like, throw

Fable at it, and it'll materialize- vice but also write comments, so because it just, like, outputs all these ratings, right? It'll give you an ELO, it'll give you whatever, but, uh, Claude can just consume that and then work on itself, right? That's the feedback it needs, and people just don't know they can prompt for an Eval. So

Yeah. I love that you said that. I think it's, like, a very, like, AGI-pilled way to, like, do it, uh, because it also allows you to update and edit that system, like, very systematically. And so as new models improve, as things change, you can really self-eval. inspect where your system is, and it'll just, like, evolve Got you in with you. every part. Uh, Exactly.

yeah, and so, b- basically, I I think it's it's, you know, how we talked about how maybe product and engineering are are merging. It's also there's, like, application layer engineering and and sort of model research are sort of merging in in those techniques that they're being shared, right? Like,

That's very true.

model people are very, very used to doing evals. Application people know, like, the the only thing you care about is, Does it fit fit the PRD? And, like, Yeah, it's like it's like a whole different

It's a whole new skill

different fish. set, I think, um, Yeah.

for sure.

I think you guys put out a pretty cool blog post on this today. It's somewhat different. It's about making the UI a lot faster, but to your point of treating it like research, uh, basically define a goal of we want stuff to load faster, and then define and, you know, question things, what can we do instead of wall clock timing? Uh, there's fun quotes in here. I think it's nice to see how the actual researchers or, uh, engineers prompted Claude.

There was stuff like, you know, we had we had 12 things. We said you have 2 weeks. It did 9 out of 12 of them in, like, 3 days or something, but fun blog post for people to research for people to read through.

Yeah. I love it when we share our actual prompts and you and, like, sometimes people are like, Oh god, there was some crazy unique ways of prompting, and it's just like, Try harder, Claude. Thank you.

Uh, no, my only feedback is it should link to the workbench, which is is no longer a thing. But I used That is that to is a good that's I used to

a good idea.

link to the workbench.

Okay. Good.

It's really good. She's like, Great You can make everything feedback, great feedback replayable.

herd. Love it.

Uh, yeah, but, uh, you were gonna share more details behind the engineering story, uh, or the the the rough story that you guys were doing.

So I think I talked a little bit earlier about, like, scale is a problem for us. Um, So it's as a problem for scale and safety, everybody. right? Like, Yeah, scale and safety. Those are, like, the two biggest things we're spending our time on. Um, and I mean, it's really interesting because most people are used to talking about scale or just, like, I don't know, compute at the, like, GPU accelerator layer and much less so at the CPU, um, you know, like, old school layer.

And that's, uh, it's it's just like as much of a problem today as the accelerator layer, which is really interesting. I see And so you you cannot

Is it?

get enough VMs, you cannot get enough sandboxes. That's primarily

exactly, that's right.

just everything being able to spin up, spin down.

With with all the resources that you have, you like, really? We got a lot of customers yeah, which is who a lot of

crazy, yeah, and and think about it, because we're using CPU across, obviously, just like running our product services, running the API, running all of our apps. We're using CPU on sandboxing, all the sandboxing that we're doing. And then just, like, all the things ever like CI, you know, like, you'd be surprised how much more code is getting written, therefore, there's so many more CI runs, therefore, you need more CPU to handle that, and so on the whole and this is not just us, this is everybody that I talk to,

I would add training, training is probably usually for sure, where you start off. yeah. So there's

There's a different bucket, right, like, I'm sure they don't commingle that much.

well, they don't commingle, although, like, CPU for Anthropic as a whole is is is like a 0 sum game, you know, like, we get our we negotiate our deals with, you know, the CSPs, and we get our CPU, and we have to allocate within Anthropic how much of that goes to all the things I just mentioned, including, um, like, research sandboxing versus product sandboxing and otherwise. And so, we really are allocating across absolutely everything, and it's just everyone in the industry right now is, uh, you Shorten used yeah, you everything.

used to be able to just, like, you know, like, scale up on demand into CPU capacity if as long as your budget allowed you to, right? And like, now it becomes something that you have to, um, capacity plan for in advance, which is, you know, like, kinda punching everyone in the face. It's brand new, and people are starting to do it, and so dealing with that means a few things, like, 1 is obviously just go try to get more CPU, cool, we're doing that as much as we can.

But 2 is we wanna be able to, at the infrastructure layer, run it more efficiently, right? Like, figure out how we can make sure that we're running our sandboxes as as efficiently as we can without compromising on things like tenant isolation or properties that we like, you know, care about for security or otherwise. But then at the application layer, how can we r- how can we run our applications and our APIs more efficiently? The thing that I'm, like, most excited about as of last week, we fully rolled out the Messages API, um, written in Rust, which is one of those, like, insane feats of, like, what you're able to do in the agentic coding era that used to be.

Like, we started talking about this, like, earlier this year, where, like, we're gonna hit a CPU wall. There's stuff we need to do, and and we've done it, right, like, we've gone multi-region, we've done all these things we need to do. But we were, like, we're running this Python monolith, and it's, like it's like very CPU inefficient. Um, and so what are we gonna do to make the system

Port Python.

more efficient? Yeah. So we chose Rust, which was a crazy decision at the time because we had a team of people like, down paper, it's obvious, but our our team was like, we do not Not rust have a single people. Yeah, we don't have a single Rust developer in Okay, it's the room. just like, put it Um, exactly. Literally. And and not only that, but yes, Claude is gonna write the code. The humans, like, design the system, right?

So, like, the humans are very involved in making sure at the end of the day, it's just like a dis- it's a distributed systems architecture, right? Like, and Claude can help you both with actually writing the code, but also helping those engineers, like, you know, bounce around ideas and learn what they need to learn to design that system. But what's most interesting is the 2 things that mattered most were the plan to do the thing and then the verification of the thing.

Cause the actual implementation was, like, Claude writing code. And so what we ended up doing was we literally built, like, an entire fork of the entire API in Rust, and then started running shadow traffic through it and verifying that the outputs that we were getting from the Python system and the Rust system were This is a the same. Feature

good side though.

flags is all you need. Yeah. Literally. Um, Load balancing and it's just it's feature flags. just, I Yeah.

don't know, I never in my career would have had a team and say, like, They did Yeah, yeah, cool, Rust. we're just gonna,

Wait, do you do you like, support still have Rust the entire people or no no Rust people?

Uh, well, they would probably call themselves Rust They're rusty people now.

people. I mean, there should be no such thing as Rust.

Um, and along the way, we have hired a couple of people who have more experience Sanity with Rust, but check. yeah, but it it still is just like you never would have signed up for that. It would have taken years and years and years. Like, you wouldn't do it all in one shot. You would have carved out pieces of the system and say, like, Migrate this, migrate that, migrate yeah, easily that. But, a multi-year like, yeah, but you would effort.

never, like, kind of one in one swoop, right? Like, port the whole thing over and then just, like, slowly turn it on, but that is what we ended up doing, um, and it worked. And so, not only did we actually save a lot of CPU, but we, uh, our latency has, uh, seen a nice improvement, which is pretty cool too. Um, and, you know, it's just it's it's a crazy era we live in cause you just never could have done that sort of thing previously.

Uh, order of magnitude of savings on the on the CPU side? Ju- just for people considering this.

yeah, yeah, I'm, uh, okay, order of magnitude, that's really interesting. Like So

Ballpark. Yeah.

Yeah, ballpark. I guess I would say, like, we've sit somewhere like 40 to 50%, something like really crazy. And then latency, I think we saw somewhere between like 10 and 20%, um, improvement on like time to first token specifically. So, a lot of just really cool improvements, uh, across Like, higher the board.

on, like, uh, longer running wo- workloads. But so it'll it'll roll out, um, soon. But it'll get faster.

Do you fight the the urge? You know, I've been involved in some major rewrites. There's always like, What if we add this new thing that didn't exist before in the rewrite? Since we're rewriting anyway, right? Uh, but then, like, that adds scope. So do you just, like, avoid that? Is So we it because of pure rewrite?

actually we did. It was a pure it was, um, maybe the thing that I would call it is like a port. What we did was we we did exactly that. We, like, resisted the urge on all these things. Then, uh, and because what the team already had to keep up with was little, like, swap team that we put together to go and do this was doing this while the whole rest of our team was, like, building features into the API, right?

And they did all the, like, static analysis stuff. They would yell at you because you built a feature over here and you didn't consider Rust over there, right? And, like, they dealt with all of that, but they resisted the urge to add features themselves or, like, make the system different. Now their next goal is to go and do some sort of rearchitecture type work, including stuff that I think will be more user facing. Like, we're starting to roll out some things that are like, you know, what if we were able to not have you have to, like, pass back and forth the entire, like, conversation state, right?

And, um, if you didn't have to do that on the Messages API because you were open to, like, you know, uh, storage of that state for some period of time, right? Then, like, especially for long conversations where the context really builds up, like, you get a lot of really great latency and efficiency wins there too. Like, these are the sorts of things that the team that was doing that port was like, Ooh, Ooh, they could like, what if do we do this right?

And Doesn't your caching I'm like, not yet.

API have that?

So we have prompt caching, yeah, but you still have to, like, pass back the state of the conversation. very, very Yeah.

niche small subsets, but I guess this is the level of No, this is detail they went to.

this is this is important. Uh, yeah, yeah, yeah. Uh, there's a lot of savings they can be done, Yeah, from a but you also turn sw- from them into a cloud Yeah, and then from cost.

from type of from number of users point of view, it's small. But if you look at the actual token volume on those those types of workloads, it's, like, going to be much What I'm ve- like, larger.

like, what Cloudflare do with durable objects? I I I feel like every cloud provider should clone. Do you do you know what I mean? Like, right, this isn't that the right shape for a messages API that needs to continuation?

Yeah. I mean, it is really interesting cause I think what we're we are experimenting with various, like, what is the right shape of, like, how we should run our compute in order to power all these sorts of things. Um, but it's also interesting because we can't necessarily rely on, like, because we get our compute across we run on all 3 clouds, we're running all over the place. Like, we don't necessarily have, like, 1 shape of here's exactly, like, the compute the types of machines that we know we're gonna run on and the way that some certain compute works that we can, like, rely on across the board, if that makes sense, other than, like, the most vanilla versions of these

things. But I think these things are all gonna, I don't know, continue to innovate and evolve as time goes on, and so we'll be able to take more advantage of those things.

Yeah. Uh, I'll just mention, if people want, uh, open-source, Cloudflare durable objects, uh, Rivet is the open-source equivalent. Um, there are 50 people listening that will kill me if I don't ask this question. You want CPU, they have CPU to sell. Who do I sell to? Who do I email? So Uh, and what do you what, uh, you know? And, like,

Open source.

what what is a what is a good help them help you, right? Like, what what is what is good look like?

Well, I guess, like, uh, email me

Are you sure? at like I don't know. They're you're probably

gonna get a lot of No, no, but gonna I'm support calls. like

gonna get a lot of emails.

but, like, you're you're looking, like, minimum fleet size, SOC 2, obviously, enterprise, like, yeah, like all given all the privacy

And for what it's worth, guarantees. like, so we have a really awesome team that secures our accelerator compute, um, that is now spending a lot more time with us on CPU and, like, where and how we can procure it, um, to make sure that we can just continue to run our systems at the scale that we need to. And so it's not actually, like, my job, thankfully, is not to go and and run around try to, um, procure compute, but we do have teams that are doing that, which is really great and really important.

But, obviously, like, there's a quality bar that matters to us, there's, like, a scale bar that matters to us. We're not gonna,

Yeah, of course.

like, go and integrate and run on certain machines if, like, we can't get enough of it. And so, all the properties that you might expect of a system at this sort of scale is all what matters to us.

Then the related to the compute problem is the sandboxing problem. Uh, people have trouble containing agents in sandboxes. They keep breaking out. Uh, what's going on there? What's the state of the art?

So, state of the breakout recently?

Yeah. It's interesting. One of the things that we did with Claude Manage Agents specifically is the way that we architected it. We put out this blog post at the time of launching where we talked about we have the brain separate from the hands. And what that means is, like, the harness runs on a durable server, really ours, right? And then

Oh, wow. Hey. Nice, nice talk, nice talk.

that's good.

Yeah, that's good.

Um, and the hands are it like a execution environment, Perfect. like a sandbox or some sort of container, right? The part of the reason why we did this was for, like, latency and reliability in these sorts of things, right? Like, if the harness is just running on a durable server and it's only spawning a sandbox or a container to execute when it needs it, that means we can get you tokens back really quickly on your initial request.

It means that your sandbox can die, but, like, your whole agent doesn't die because the harness is durable and over here and the har- you know, the agent's storing state separately also. And there was a whole product development journey that we went on to get there because our 1st iteration of an API for agents was just, like, throw Claude code in a sandbox and, like, shove an API on top. And the reason why we didn't launch that as our, like, durable, scalable solution is everything that I just said.

It was like, you gotta boot a sandbox and boot Claude code before you can even get a token back, right? The sandbox dies, and you lose your whole agent, and so reliability wasn't where we wanted it to be. So, we architected this way mostly for reliability and latency. But the interesting thing, there's actually a a the product manager, uh, who worked on this from the very beginning was, like, the most important property of this architecture is security.

And everyone was like, Yes, yes, yes. But latency and reliability. And, Now, like, like, now, now they're That's the only never gonna get I literally way. I apologized to this person yesterday, I was like, I'm now, like, telling

I understand.

everybody who I meet about what you've been saying for months that, um, nobody was was as excited about as as this person was. But one of the more important properties of this architecture is you have credentials for any any work that the agent needs to execute, right? Like, if you're gonna go talk to an MCP server, or if you're gonna go, like, use a CLI or call some API, whatever you might wanna do, right? And, like, you need to authenticate those things.

You need to store those credentials somewhere so that the agent can use them. And the way that we architected this was we have a totally separate vault where those credentials are getting stored. And as once the agent decides, like, I'm gonna go and call out to one of these services, the agent only ever sees, like, a placeholder, right? And, like, on egress, we have a proxy that goes and, like, pulls the right credential out and injects it and lets that out.

And so what's interesting about that is, like, the agent is never, like, in the same place where the credentials are, right? Because in the old world of, like, just boot the agent in a box and, like, let it run, you had to send credentials in there. And then you're, like, doing some sort of, like, Ah, how do we, like, scrub this and that and make sure that, like, the agent can't take that credential and go do something that it shouldn't be doing?

It's just, like, literally a, like, literal infrastructure wall that says the agents, like, can't access those credentials and can't do anything bad with them. So that was one thing that was really great. The other nice thing is, like, on that sandboxing infrastructure, we can make sure that we have really great egress controls. So we have, like, domain allow lists and things like that, or you just, like, totally shut off network access. So that you can't say, Okay, agent wrote untrusted code, that code executes in sandbox, and that code can, like, reach out to the internet, right?

Like, now it only can reach out to where you say it can reach out to. So, there's a lot of really great properties around all of this within the architecture and within the way that you can configure your sandboxes that make it so that you end up, like, secure by defaults. One of the the things that's starting to happen is as we're talking to customers who are like, I wanna run agents either for, you know, a product for my customers or internally.

But I'm, like, you know, hearing about incidents where agents are breaking out of sandboxes and bad things are happening. How do I think about these things? I think the 2 most important things are architecture, so that it's, like, literally impossible for an agent to do something that isn't okay, and the configuration that you have on those sandboxes. And so one of the really, really nice things about cloud managed agents is you just, like, get that out of the box, and you get something out of the box that, like, as we learn more and we continue to evolve the properties of those systems, you kind of get that for free.

Cause I think, say, all the words that just came out of my mouth, like brain and hands and credentials injection and egress and all these sorts of things are just, like, if you go back, like, 9 months, like, no one would be saying these words, right? Like, these are all, like, relatively new problems that we're solving, and so, um, we want our customers to be able to take advantage of the solutions to those problems as we're even learning about them and solving them ourselves within our infrastructure.

Yeah, and internally, we're just upgrading everything to be, basically That pattern, secure by default, yes, but it's better.

uh, the we we also unlocks, uh, something that we talk in the Thadric pod, which is the Cloud Brain local hands, which is products. Is that Jax? Is it the rough shape of what's going on?

yep. That seems right.

Yeah, I I f- I feel like, uh, it's weird because obviously you're famous or Cloudflare is famous for being the super unsafe, uh, Dangerously Allow Permissions thing on on your machine. But now you're, like, finally evolving to where you probably should have started in the first place. Basically, I think, like, all these sandbox leaks, right, are they, uh, a fundamental issue with our s our understanding of sandboxing or or container tech, or are we just bad at this?

Like, are we are we just, like, not disabling like, not doing the allow list right? Like, it sounds like we're just not doing the allow list.

Yeah, so maybe a distinction that I would make is in a lot of the incidents that people are hearing about and reading about, these are incidents that are happening to models, not just our, like, everybody's models, right, that are in training. Where the way that you're running these models is you, like, put them in a box and you say, like, Solve impossible task, I'm gonna stress you out. Ready, set, go, right? And you, like, can only do what you can do within your environment, and the agent is basically incentivized to try to get out of this box and go solve the problem.

When people are running production workloads on our API or on anybody else's API, you're running in an environment where, like, there are safeguards that are running that are, like, not necessarily running in training where, like, the whole point is to actually train the model, right? But in production workloads, first of all, you have those online safeguards, which are really great. Second of all, you have more of this architecture and infrastructure around what's happening. Like, the agent isn't just in a box trying to get out, right?

Like, the harness is over here, the hands are over there, and, um, it's just, like, not even a problem that you would have to contemplate. And so I think that some of this is just, like, people having the right understanding that what's happening, some of these incidents that are happening in model training and otherwise are just things that, like, wouldn't happen in production workloads, where a lot of work has been done to put the right guardrails in place so that those things, like, never really could happen.

The 2nd thing I would say is just, like, everybody's figuring all of this out, like, relatively newly because, you know, long-running agents is a relatively nascent thing that everybody is building and trying to deal with. And so everyone's figuring out some of these things on the fly and just racing to put the right, you know, like, controls in place. And, you know, some of that is just happening as we go. So I would say it's not, you know, we're people are bad at this, right?

I think it's just more it's a nascent field, people are figuring it out. And I think when we talk about things like pacing the frontier, the way that we think about it is really specifically, we're doing all the work that we're talking about right here for all of our agents internally and making sure that all these things are accessible to all of our customers. So that, like, by default, the right guardrails are just in place so that you don't end up having agents doing things that you don't want them to be doing.

Fair enough. Um, uh, may- well, cause you mentioned pacing. Uh, we gotta go there. We also talked about this last last time. It is, uh, superficially, uh, contradictory to advocate for pacing and then also release a new frontier model 11 days later. To me, it's just it just means that it just passed your safety checks, right? Yeah, and and that Uh, and

Yeah, everything we talked about earlier on, like, how we launched the models, um, it's like we are continuing to evolve more and more and more safety properties in place again across these, like, I kind of am starting to think about this in these 3 dimensions. It's like the actual, like, architecture and infrastructure security, then it's the online safeguards that are running when a model is doing its work, and then it's the offline safeguards classification that's happening on data at rest.

So across all those 3 dimensions, we're putting all the right work in place to make sure that we feel really good that we can put a model out there, and it's going to be safe. And really, that's fundamentally what we mean. And so for Opus 55, it's an excellent model. It's also really, really great on all the dimensions that we care about across those 3 things, and therefore we feel good about releasing that model. So, we went for what we thought what we were hoping is the best possible user experience within this while we continue to hill climb on making those false positives better.

But it's obviously like it's still a frustrating experience for people when you're trying to work with a model and then you fall back to a different model. And so it's just something we have I I think to keep working on.

people also understand it in the safety sense. Like, you want more false positives, right? Like, it's a safety issue. We clearly see the whole Venecias incident, and you're still giving access. Yeah. T Yeah.

t to me, this is like a a a close cousin of the general model routing problem of, Do you want a smart router or a dumb router? Does that make sense? Like, uh, because as far as it can tell, the the the Anthropic fallback router or, uh, detection thing is a separate model that's online. Uh, it's So

it once it hits a classifier, there's just logic to what you can stop, yeah.

right. Like, so it is not the full fable, right? You're not you're not s you're not, you know, running full fable at it. And that that that's gonna cause classification issues cause it's not it's clearly not as smart as fable, like, clearly. Um, so then there's the question of, like, well, then, like, when I have a classification issue, that's that's like an unjustified fallback. Like, you never you never need a fallback, I'm just doing my bio homework.

But I mentioned bio one time, and then it kept on falling back. Uh, so, like, there's a question. Th- then this is very similar to the routing problem because,

It's very similar.

like, you're you're routing models, and like, how can this the the a dumb model route to a smarter model if it doesn't even recognize this? That's that's a smart model problem. Um, how do we solve this?

It's a hard problem.

I I think I think in some super alignment sense, models are good at classifying stuff, but not answering them, right? Uh, uh, not the smartest model can tell this is a dangerous query, even if it can't make any impact on that. So

I think the super simplifying thing to say would be like, basically, you have the smarter model figured out over the one that, um,

What?

it needs to detect. I I think it It's lower also

x cost. There there's trade-offs

There's trade-offs. there. I'm not saying it's, like, the right user experience, but it would probably be a slightly more I I effective route.

think one thing interesting is you also don't cut off immediately. I've seen cases where it will start answering, start reasoning, and then get hit by a safety filter. So, w- my 1st example with Fable was throwing the Fable safety card. And it started re I was like, Just break down anything interesting in here. Starts breaking it down 5 minutes in, And so I think it gets into cybersecurity, and it's like, Oh, we're cutting, we're rolling back to Opus.

But it did a lot, there's a lot of context and I see it. I it got to a point where it's all about cybersecurity safety. And then it rolled back, and it's like, it wasn't rolling back my query. It was the combination of reasoning and the whole chain, right? Do you have a idea of Yeah, less brainstorm, what's better?

what what what's better?

I think the user experience and how we resolve that user experience is probably, you know, a separate question. Like, maybe it's a use a super smart model to handle it or or whatever. Um, but I think that the more appropriate UX that I would imagine for a user is to actually have a more consistent, unified experience with Claude, where instead of getting you falling back and then you can double fall back, there are, like, logical points, um, in our tree that that do do that for, again, like, safety reasons and whatever.

But, I think that UX is just, like, really convoluted for a user. It's like you almost you tr- you're trying to do something, and I I would probably take the case of, like, a good user, and you're a good user, just trying to do something, solve your bio homework, like, assess the cybersecurity, like, safety card, totally reasonable. And you get blocked and you fall back, and I think instead it would just be like, you are well, our model should have gotten or our classifiers, or however we wanna do this, would be smart enough to detect that that is a situation that happens and just write down the me the answer for you more concretely while dropping down the nature of

the, like, uh, negative use of that particular capability. So going back to that bio homework scenario, while we may have classifiers that have detected that it is possibly

I'll also say, like, the f- on the API, we went with a spectrum of solutions for this for customers. So, first of all, like, if you've built something on the API and you run into a block, like, you don't have to, like, continue the conversation on a fallback. That's what we've done in our apps, and we've made that a feature of the API, and you can do it, like, server-side within our API. You can choose to, like, set, I want to do the fallbacks, right?

And, like, we'll do that. We also have it within our SDKs on the client side. There's helpers for if you want to, like, manage fallbacks in various ways, or you could choose to, like, re-prompt differently in some way, right, and and continue with the same model. Um, but I do think that there's some in-between interesting solutions that we'll continue to experiment with to figure out how we unblock people and and do so safely, but with a good user experience.

Yeah. And then you brought up the the bit about it's, like, very similar to routers, I think, like, I totally agree with that. I think it has a similar UX problem as well. Like, with the router, you you go through, it's like, let's say you wanna do a a cost router, right, for the task complexity that you have. Makes sense. Uh, very hard to do. Uh, you don't know everything upfront about your task, so you can't, like, even with perfect you just don't actually have perfect information.

Even if you could actually understand each model perfectly, which you don't, but even if you did, you also don't have perfect information about your task. And so there's, like, a lot of error associated with your your routing logic that you would have to create, whether no matter how good a classifier you produce. And yeah, maybe it's, like, a super big brain model type of problem, but regardless, it's, I think, a very difficult one. And in each one of those, like, scenarios, if, um, to find a way to, like, route that properly, it's very, very convoluted.

You could expose all that logic to customers like, Hey, you know, your task was to center a div, so I'm gonna route you to something super cheap. Uh, and like, then this other one's, like, detect vulnerabilities, so I'm gonna route you to something, like, really expensive. Um, if you expose all of that, I don't think that's, like, a great user experience ultimately either. And for that similar class of problem, I think the more idealized UX, uh, would be something along the lines of, like, dynamic, like, effort or dynamic, like, understanding of the level of complexity of your task, which you might actually evolve over time as you learn more and more information about the task.

But in that world, if you're kinda, like, moving between a lot of different models, it actually becomes very, very disjointed. And so, like, maybe a more unified approach would be something like it's a consistent model, but on it's, like, very, very low effort level, or something like that. It's actually very, very cheap and very, very affordable, but it can scale itself up very, very continuously, but you have, like, a very, very consistent persona towards that model.

And obviously, speaking about this in the context of, like, you know, like, our family of of models, if you try to even do that across other family of models, other model providers, like, the personality, the disjointedness, like, starts becoming very, very complicated. I don't know how you solve it in that layer, but at least within, like, our family of stuff, like, I'm more bullish on things that are just more consistent, and those might be ways to solve it, uh, more robustly for customers.

Yeah. You're you're not responsible for anyone else, but within Claude, yes, yeah, I mean, that this is your your territory.

It's nice to see the way you guys think of these problems, right? Intent, in and of itself, is a hard problem. Like, I feel like the models have gotten better, but sometimes people don't do a good job at classifying what they want, or they just slider all the way to one extreme. Or, you know, the model doesn't really understand. You're asking a simple question, and it's like, Oh, I have a whole HTML landing page now. I just expected a few line output, you know?

So there's always a a mess around there. I

I think we we in in this conversation, where we talked a lot about safety, we talked about, like, watching the models, like, all this kind of stuff. But it's obviously in in pursuit of helping power the workloads for people. And, like we discussed, the kind of the earlier part of this conversation, we are seeing workloads, uh, change quite a bit. I think, like, in some workloads, intelligence does seem to be increasingly saturated, which is, like, a a new, uh, situation.

Um, and it's it's getting to the place where I think there are some class of tasks that are still very, very unsaturated. And actually think they're more than maybe people anticipate, but they're increasingly going into the direction of you get that alpha, um, or that return by doing extremely long-running agentic tasks on things that historically you were not really able to brute-force, but maybe you can with a lot more intelligence behind it. Um, and those workloads are really, really fascinating to us, and we see a lot of value in that kinda area, both from businesses getting that value from themselves, as well as, you know, like, new companies or startups or however you wanna frame it, um, who are interested

in building kinda, like, agentic products on top of it. I think if you look at the class of things today, a lot of it tends to be it's obviously a lot of coding stuff. There's obviously a lot of, like, enterprise, uh, innovation, acceleration kind of stuff, um, but they tend to take the shape of automating a very distinct and specific process. And I think in some areas, that's actually really starting to work, which is awesome. In some areas, it still isn't, so there's still a lot of, like, room to to kind of evolve and push the frontier for those models.

And then there's, like, this new class of workload, like the the bio stuff that we talked about or some of the cyber stuff. Like, that's a it's just a very, very complicated set of, like, I would say, hard engineering or hard science problem. Um, and those workloads are increasingly sharing up as well for us as, like, a really interesting space to just deliver value to customers.

You know, you you are at one of the few spots in the world where you're discovering all these workloads, and we're automating it. It's like,

Wow. Yeah. It's pretty crazy.

Uh, I mean, you know, that's, um, very inspiring. Uh, you guys are having so much impact, and and I'm really, uh, grateful for that. We're gonna go into a few closing questions, which I which I thought would be fun to sort of bounce back and forth between the 2 of you. Um, how do you fight slop on your teams?

So, we actually this has been an interesting debate within our team. The the we're no different than anybody else, right? I think plenty of teams are talking about this. Um, and what's interesting is what we're trying to figure out is, if you're producing a document, you're producing something that you want other people to consume, code, documents, whatever, right? If like, you should use Claude to help you, of course, right? But if you use Claude to, like, totally generate a thing you didn't spend any time on, like, actually reading it yourself and making sure that it was something of high quality, then you're putting a whole extra burden on the consumer of that document or of that code or whatever to

spend a whole bunch of time trying to parse it. And that's just, like, insane and totally unfair. But what's funny is we, you know, we push back we push on our team, like, You can't do that, right? Like, If you're producing something, use Claude, sure, whatever, but, like, you have to make sure that it's a high quality thing. But then we You as the human get into these

are still responsible for the quality.

Right. You're still responsible for the quality. But then you get into this really funny thing where because of the Claude-isms, even just, like, silly, like, I don't know, there's, like, an em dash in the middle of the thing, right? Like, people's brains trigger on this, like, Oh, no, I'm reading something that was written by Claude and not a human, and I'm, like, annoyed because I wanna be wr- like, reading a thing that was, like, fully produced by not necessarily fully produced by a human, but, like, that has the quality bar that the human would hold, right?

And so then we get into funny situations where maybe actually, like, the person did spend a lot of time making sure the thing was high quality, but they used Claude to help them, but the there's those little hints of Claude, and then the reader is, like, turned off, you know? Like, that's, like, unproductive as well. And so, what we have kind of tried to do is, like, set a bar within our team that's very much like, Use Claude to help you, because it would be really silly if you didn't use this hilariously powerful agent to speed you up in your job and in your life, or to just, like, help you produce better things.

But it has to, like, have a certain bar. And then for the consumer and the reader, if, like, it really actually is just slop, like, send it back, right? Like, just say no, just send it back. Like, you're not gonna waste your time on this. But also, don't be, like, too nitpicky about the little Claude-isms, right? Because, like, we have to find our right balance here. And so, it's just really funny because we're, you know, we're going back and forth on this stuff within our teams and just, like, so is everybody else in the world right now, um, we're just not any different.

But but you have free tokens. Uh, well, sure, but, So

um, I mean, it that doesn't tend to matter quite as much when it comes Yeah, as to, like,

it as in, so you have more than anyone You probably else in the world.

see a lot uh, you guys use a lot, you see a lot more people doing a lot more.

Yeah. Yeah, it's totally fair, but it's definitely it's definitely

Heavily discouraged.

a yeah.

Uh, engineering slop?

Yeah, I mean, it's same thing,

It's a big one. right? Like, but Okay.

but I do think that where where we've gotten to is a pretty good spot where, like, you know, humans play a large role in making sure, like, designs of systems are good, right? And, like, implementation tends to be Claude, right? And, like, we're in a pretty good spot where for, like, a lot of the code that gets written, like, all you really need is just, like, a quick stamp, and you can keep going. And, like, you don't necessarily have someone, like, deeply reviewing every single line of the code cause you care more about, like, the overall plan, the overall, like, design is in good shape, less so than, like, the exact perfect implementation.

And obviously, rely on a whole bunch of verification to make sure the thing that you're gonna ship is high quality. And so it's actually been, like, less of a conversation these days on, like, code quality being slop, in part because Claude's just, like, gotten better, but in part because we have a lot of the right verification in place to make sure that, like, you know, we're holding a high

I see.

bar. Yeah.

I see. Trust your tests. Um

Any product PMing slop, you can ship features a lot, you can put stuff out in a uh, so he day.

said this to me recently. I did. Do you remember what you Uh, Tharik said to me?

Tharik, who did this? The right way to use model capabilities is not to ship 10x more features into prod. Uh, spend more time understanding your users, trying experiments, building whatnot. But, basically, you know, you could very easily just keep shipping random slop, but you you should actually think about it. You should spend more time kicking off that first prompt. I know now today, he's like, Kick off plan mode. We don't need it anymore. But, uh, outside of plan mode, you should think about what you're doing, right?

It's very easy to put out a lot, and you can get product slop.

Yeah, I mean, I I tend to kind of go on this rant about there's a lot of just, like, fundamentals about how software gets, like and products get planned and shipped and otherwise that, like, doesn't need to change just because agents are, like, helping speed up certain parts of the process. But there's this inherent, like, almost laziness that's starting to come out from a lot of people where it's like, Oh, I don't have to, like, think about some part of the process at all because the agents are doing it and helping me or whatever.

But, like, those are still, like, incredibly foundational parts of the process that, like, should be things that you're spending your time and energy on, and I think that's kind of the spirit of, um, what he's getting at here.

It's just like, as a founder, or C creator, let's call it as well. Um, it just feels like because the cost of experimentation or exploration is so low now that I should be doing more of that. And he's basically saying to not do that, to to, like, use it in service of a a very intentional goal that I start from the start. Which, like, uh, I obviously, I'm coming around to. It's just hard. It's like you have this toy that does everything.

It's just like, Oh, just throw it at the wall. Yeah, there is

scope too. So, like, the blog post of today was two weeks to make Claude AI faster. There were twelve things, uh, we had there were thirteen things. We hit twelve of the thirteen targets in three days, right? Well, you could just stay, Okay, it's a two-week sprint. We'll do the thirteen. But instead, there's twenty things you push bound. This is not product slop, it's not new products, it's refactoring. But still, you guys do a very good job at, hope, putting a lot but keeping high bar quality.

Okay, part 2 of 3. Uh, this is gonna make Angela very uncomfortable. This is Angela Tweak Corner. Uh, I will tell I, I will I will, uh, I will tell people nonstop Oof. to uh, I and I I I'm pretty sure you've deleted stuff because I'm like, there's there's stuff that there's stuff in your deletions that I'm like,

That you like.

Still, yeah, I know. They're like, really, like, alter my DNA. Oh, I appreciate Well, uh, that. Wow, what do you refer

that's a nice compliment.

to? What do you think about the most? Uh, what do you refer people the most that you wanna sort of bring up? And I can I can pull it up for you if you want, but

Um, I actually I don't think I have a favorite, honestly.

no, come on.

Um, no, I really d and okay, I guess maybe I like the one that you say where it's half doing something is expensive way of not doing it. Maybe Yeah, that's that's my favorite.

that's very on topic for us today.

Yeah, I guess. It's Yeah, okay. There you go. it yeah, it's

But I I guess, uh, y- you know, you you obviously, you're in a different job now and you don't do this anymore, but, like, uh, that I I think this is a power tweet. This is like a Na- Naval Robinson. It's Naval versus You that do this and, like, uh, what

I said, stopped. mode what mode But yeah.

of what, like, what framework of mine frame of mind Frame of are you in? mind. That's a good

Six years ago.

question. Yeah, no. So, this is a very long time ago. Um, I went through a period of my life where I was, like, reading constantly, actually. I found myself, like, very bored, um, of the stuff that I was doing, and I felt like I was just, like, not being, like, intellectually challenged. And so I actually went on this a bit of a bender where I I think I read something like 200 or so books in, like, a year.

Um, and I was just curious about, like, every I just wanted to, like, read everything, and I was missing a little bit of, like, the the kind of, like, intellectual, like, I don't know, like, muscle building you get when you're, like, in school, where they, like, you, like This is a new topic. You gotta learn. It's, like, tough. It's, like, not super fun, actually, but you are really, like, stretching yourself. And I felt like after just, like, working for a couple years, I just, like, stopped doing that.

Like, yeah, the job's, like, interesting. I guess it's, like, challenging, but it doesn't have, like, that, like, intellectual bit to it. So, anyway, I went on this bit of a a bender, uh, a reading bender, uh, and I, like, read, like, 200 books in a year. It was great. I had, like, a really long commute, so it was, like, it was a good use of my time. Very long, like, car train ride. But, uh, I ended up coming out of that and being, like I was just, like, thinking a lot, um, as a result of that.

And at the time, it really wasn't to build, like, a Twitter following or anything. I was just, like, tweeting into the, like, abyss. There was, like I had, like, no followers. These were, like, tweets for myself. And I was, like, practicing a little bit of, like, how would I try to compress the idea that I have in as few words as possible? It was, kind of, like, the personal test I had for myself. And then, I guess people just started liking it.

And so I just I just kinda just kept going at it. And then I got too busy with work and I I stopped doing it. So

I gotta ask, after all the reading, any book book recommendations, blogs, long articles?

my favorite one still is probably The Black Swan by, uh, Nassim

Must you tell it?

Taleb. Um, that was like one of those, like I was, like, to me at least, it was like a bit of a like a life-changing philosophical book where I was like, that's a completely different way to look at the world. And You know, it caused it.

thinking of Black Swan when the,

Oh, really?

you know, the fable thing happens, right? It's a thing that you didn't know Very much could happen. Now that so. it happens, like, it's a it permanently changed your framework of how things work.

And, uh, the idea is, like, so simple, but I think he did a great job of just really hitting it home and articulating it and how it's like a common property of, actually, of so much of the world. So I really love that book, and that's probably my favorite.

Uh, and for listeners, if you're bored with that one, you do Foo by randomness randomness, also by Taleb. Did you like, uh, Before he got Skinny, Fragile? Not Less so,

so much. Yeah.

less so. The the more famous he got, the more intolerable he became. Uh uh. But al- also question for Katelyn, my i if you do have one, do you have a similar, like, This is a book I recommend to everybody? No, that's a good Okay. Yeah.

question. I didn't even think about it. It is funny cause I do think we have very different styles on this front. Like, she's no longer, like, in fortune cookie tweeting mode, but

Of stuff. I do think that, Yeah.

like, you stopped that. But I do think a a big part of how you work and how you work with other people is, like, philosopher mode sometimes, Do you you know?

drop fortune cookies in Slack?

Oh, I I do actually. All Not intentionally, I think that's so much a similar R- I think that's kind of intuitive a poor success algorithm.

of succeeding in Anthropic is, like, you have to be very good on Slack. Yeah. Oh, you do

That's interesting.

have to be very good on Slack. Um, it is it's a pretty intense place, but I think I'm a little so, I'm kind of psycho about, like, inbox zero on both email and Slack. It's easier on email these days than Slack, um, but I like pretty aggressively prune, like, here are the channels that I'm, like, actually in and paying attention to versus the ones that are, like, muted below the line, sort of thing. And I'm, like, always making sure that I'm on top of the ones that I need to be on top of.

But it is it's basically it feels like a full-time job. Like, there are days where I'm like, I could cancel all my meetings and write no documents, do no work, right, and literally just sit there, like, on Slack the entire day. And I would have done what felt like an entire day's worth of work just by, like, engaging on Slack with all the things. It's pretty wild.

No, but that's horrible. What what So that's do you do?

exactly how I feel. I feel like you do all this, eh, you put in all this effort, and yes, like, you you met the mark on, like, responding to people and getting certain things over the line, but you didn't actually produce the output that you wanted to. And, like, that's, like, a that's, like, personally a very hard thing for me to deal with. So I try to be like, I need to, like, dedicate time to, like, actually thinking and actually, like, trying to produce something in order to feel like I'm, like, actually And ignore doing something.

it and drop stuff, right? You you're gonna drop stuff,

Um, yes, I have gotten a lot of feedback. I'm a little hard to reach on Slack. Um, Yeah, pretty terrible as any change thing that happens now any change

is I get a lot of Slacks saying, Can you get Angela to respond? on It's just like,

the daily. I'm like, I'm like I'm, like, really

I know.

trying to write a doc here.

Wait, wait, so you guys don't have, like, a personal Claude, like, assistant, something to I There's there's don't know to gotta be a help you? Like, limit on, but there's So I have a plenty no one, no one written in, in like, a personal

Nando's law.

EA for each of you

Yeah. Well, so we do have we have a human one also. We also have we have plenty of Claude help in Slack, for sure. Um, but I do think actually the the types of Slacks that I'm, like, when I'm pruning through and getting are often just, like, our team is, like, blocked on something cause there's some hard decision that needs to happen or whatever it is, right? And it's like, it is just like a form of, whether it's in Slack or some other way, right?

Like, it's a form of getting the team unblocked, like executing in high quality. That is my job anyway, and it, like, happens to happen a lot in Slack, but I do agree that it ends up feeling like, What did I even produce when you get through a whole day worth of literally just, like, going back and forth with people and trying to solve problems? You you produced And and

a a functioning team. You that that that's unblocked. So I will give you props to that as a as a former manager, that is Yeah, yeah, yeah. You don't feel like you could you produced anything, but, like, you produced a team. Yeah. Uh, speaking of your team, uh, I was gonna go to you next, uh, Katelyn, uh, Devrel Corner. Uh, we're I know you're really hiring devrel time. We're making a call out for for developer relations.

Uh, you hired Adi Osmani recently, Uh, great get. Uh, Lance, uh, and you know, I don't know, but you have you have you still building out the team. Uh, what, i what do you guys want to hire for? You know, this is this is it to talk about it, and I I'm happy to So trade notes.

Yeah. Well, props to you because I will say I I'm the when outsourced Anthropic recruiting, uh, you k- uh, Katelyn recruiting literally, team. It's literally. Um, and fun. it's un- it's actually not even just recruiting, like, I'm gonna, um, toot your horn a little bit. When so, I when I joined Anthropic, we you know, we're a small team. We're, like, all things considered, we're an extremely small team. And so we run the platforms, the API business, and and all of our product infrastructure.

And, um, we we had, like, a very nascent devrel function within Anthropic that was, you know, like, not necessarily within the product engineering org. And when I joined, um, we were like, Okay, we, like, need to grow up a little bit and need to figure out how to, like, engage with the community, engage with the world. And, um, we need to, like, really invest in this devrel function. And, um, we ended up spinning up a devrel team within platform and a devrel team within cloud code.

In part because it's, like, 2 pretty different, like, developer ecosystems that we're trying to engage with across the API business and agentic coding, right? And so we ended up doing it as, like, 2 separate teams, but kind of, like, dotted line to each other and spend a ton of time together. Um, but it was it was something that I was going to be responsible for, like devrel, which I've spent plenty of time with devrel before at Stripe or otherwise, but it was never something that I was responsible for.

And I was like, I have no idea what I what I'm doing. And you were like, I got you. Like, here are the 3 pillars of devrel. Like, here's how you think about it, which was really great across, like, product, content, community. And then now I can, like, rattle that off to people, and they're like, Wow, you really understand devrel. I'm like, No, so it's You're like, not me. Um,

I copied that. Yeah.

yeah, that that was not It it's the original plan.

apparently that post is apparently, like, the default reading for Google when when they onboard new devrel. It's actually, like, really funny.

Yeah, um, it's it really, like, this this really helped me because I had no idea what I was doing. Um, but then on top of that, we, you know, like, I've I don't have a manager for the devrel team within platform right now. And so that's one of the things that, um, I'm hoping to hire, and our devrel ICS, um, report to me directly. In part because I think it's a really, really, really hard problem right now, but also in part because I find it really fun and I'm having a lot of fun working with the team.

But really specifically, like, what's hard on platform is platform in some sometimes in our, like, public narrative. Like, our platform business is a really large business, but it can kinda get lost in between research and policy and all this work that we're obviously doing that's, like, our most frontier important work. And then our applications, like Claude Code and co-work and all the things happening over here. When you you think about Anthropic and you think about our public narrative, those are some of the things that often come to mind.

And less so that we're running this, like, massive developer platform business that's really important and so many people are building on top of. And so a big goal that we have with our devrel team within platform is figuring out not only, like, the really classic, like, how do we better engage with, like, you know, product, like, what people are building and how we get feedback back into the product and these sorts of things. But a lot of it's just, like, finding our voice and finding what our brand needs to be and what our narrative needs to be and, like, who we even are gonna be in the community and not kind of get lost in all of the things that

are happening. And so that, I think, has been interesting cause I think the brand that we've started to spin up for our devrel team, in part because actually Lance joining our team was you connecting me with Lance, but Lance is highly technical and the way that he goes about devrel is very much like teaching, right? Like, he writes a lot about memory and managing costs and these sorts of things, and he goes really deep. He builds evals.

He, like, has a lot of really technical content that he puts together. And I really like that idea, like, that idea of our our devrel team being kind of, like, thought leaders for people within what is a really nascent, really hard, really evolving space.

Educators, thought leaders, explainers of things. Um, and there's another version of this I really like called the Master Builder, uh, which is I will build the things with the platform to show you what's possible. And then you'll be like, Oh, I didn't know that was possible with with the Claude Platform. And then you j- you just go. Uh, and this is a thing we've stolen from Lego. Uh, yeah. Because Lego, how do they sell Legos? They have master builders.

They build things with Legos, and you're like, Well, I wanna that's the do I wanna do world's coolest that? I wanna be like this dad. person. And then they they they do that, Nice, that's and so awesome. Yeah. And, um, yeah.

the, uh, CJ, the other person on our team who's really great as well, I think he also does a lot more of that sort of, like, um, you know, writing cookbooks and, like, putting out examples and really engaging with the community Yeah. Cookbooks on X.

is funny, right? Like, when APIs launched in 2023, as Cloudflare LLMs, very important. Now,

Not so much.

I don't know.

Yeah. Well, it's interesting cause you're less in the spot where you'd, like, copy paste the code and, like, use it, right? Um, but I do think it's like it it shows the art of the possible for people of just, like, what even is the use case, right? And, like, how can you go about accomplishing it? But I hear you that, like, you know, the code itself is maybe less. Like, we for the longest time had this, like, there's there's an engineer on our team who wrote our cookbook on, uh, computer use, and it, like, haunts him to this day because it was, like, massively complex and, like, you know, kind of a beast of a thing that, um, I think,

you know, a lot of people got use out of at a time when it's like, I'm gonna build a computer use implementation, and I'm gonna deal with all the the screenshotting and all the things, right, that are happening between the Messages Yeah, we were we're API. Um,

here for the hackathon that you guys launched at Novus. Yeah.

um, and, like, the that was the era when you, like, needed to have that sort of thing. And now, um, you know, it's much more about we have a skill, like, the Claude API skill. It's, like, really, really excellent at helping you either within Claude Code or any other coding agents, um, to build a really high quality integration. And now, like, these sorts of things, like, I'm gonna build a computer use integration. Like, how do I go about doing that?

Like, very few people, I I think, slash hope, are reaching for the big gnarly cookbook, and more so, they're loading up the skill in a coding agent and working Yeah. Or with that.

they're just putting it into a reference architecture. It actually do does much better. Like, increasingly, we're seeing that for a lot of the things that you might want to have different implementations on, uh, typically at the harness layer, but there could be other variants. You just point Claude at it, and you say, This is the reference architecture I want, and then you build it.

Fun fact, the Claude API skill that helps people build against the API is owned by our devrel team. Um, so As I was they should built by Lance,

as they should. Yeah. right? Exactly, like, This is the most,

that's the sort of thing.

uh, high-leverage form of devrel is to, like, distill everything that is the best practices that's gonna go into those agents. One of the reasons it's weird because, like, uh, I'm regarded as this, like, devrel person or authority. But I like, it's this, like, there's some part of me that just wants to tell you you don't need devrel. Get so good Um

you don't need it.

ye- right, uh, or right, you turn the engineers, the the the leaders, um, into the public faces. Because all you really are trying to do is public faces and, like, here's how we do things here. And then it that that's kind of it. Like, you you're

Some something this natural.

anthropic, you don't need, like, you know, to you people will hear about you. So, like, there I think to me, when Boris came burst onto the scene, like, it seemed like a very, like, Hey, I'm Boris, I'm the creator of Claude Code, blah blah blah, it's how we do things, blah blah blah. And then you you're like, he yes, he has the day job, but he's also doing that func- functionally. So, like, why why doesn't everyone just do Yeah.

I that?

mean, it's really interesting. And there are teams that have done this really well. Um, like, I think the Cloudflare team is a really great They're all example of, like, yeah, all they're all You know, I'll do all people they're it for all super are

the video.

suck nonstop.

awesome, right? And Yeah.

to me, that's a actually a surprise. There's a very vis- visual sign of a good manager, is that they're all happy with their jobs, they're all they all, like, s like, publicly identifying. They're not quitting. They're not another quitting.

It's interesting because, uh, I think this is actually something we are trying to do more of with the platform team and with our team and and folks who, you know, um, can kinda just get out there and, like, be part of the community and and be speaking and be on X and be wherever they are, right? I think the problem for us is just, like, having time and kind of knowing the right way to do that stuff at high quality.

So I th- I see the devrel team in some ways as just being, like, the experts in doing this sort of stuff at quality. They can do it all day, every day, and they can help enable the rest of the team to think about how to do this with the small amount of time that they are able to spend on this sort of stuff to do it at high quality too.

Yeah, so, like, uh, yeah, so, th- I call this, like, kind of the unbundling of Devrel, uh, into a front office and a back office. So, like, a a typical person who does this job does the whole thing, start to finish. They're kinda autonomous. But, actually, you can sort of build out this back office that lets your product leads and your engineering leads, um, actually take the spotlight with, uh, a simple contract of like, Okay, you sit down with me for 30 to 60 minutes, and we'll do the brain dump, and then you're done.

You go back to work. Uh, and then we'll, like we'll, like, ghost ghostwrite for you, whatever, uh, but, like, that team of, like, production people, writing people, whatever, is, uh, sort of provides the service to the rest of the rest of people. And, uh, I I think that to me, that's, like, kind of the future of of the of this stuff, to be honest. Uh, because people, like, don't actually really wanna hear from Devrel. They wanna hear from Vance.

Uh, and and, like, you know, y- you know, and but and you have, like, lots of the Tariq you have Tariq, uh, eh, you know, and Why are you using Lydia you not that if got Lydia

it's not?

through Bun, you know? E- e- exactly. But, like, he he, like, functionally He really is correct he does the

tough Friday.

job. He does the job. Um, and so, like, that you want more of those cause people people wanna hear from people in the product. Not not, like, who are, like, paid spok- paid shills, like paid paid spokespersons. And that's Get re re which I get a lot of, and that is fine. That's that's a job somebody has to do it. Like, something like, reasonably, you can expect people who are, uh, working on engineering to not be good at it.

So, like, you should pay someone who specializes in it. Like, specialty is is a skill. But if I think there is a back office, front office separation that that can, em- emerge.

Yeah. That is really cool.

Yeah, and and, uh, my analogy would be, like, everyone hires, but you have an HR department. HR department enables

That's true.

you to hire. Indeed. Do you hire? Like, Yeah.

Makes a ton of sense.

Uh, cool. And that that was my soapbox. Uh, that uh, you know, closing questions. You've been very generous with your time. Any parting thoughts or calls to action? Use Wrote something Claude. cool on Opus

Is that what what's something like, something people should try to push the bounds with new Opus 55? You guys have been using it for the longest.

Yeah. I mean, it's really cost efficient. Um, it's less verbose, all the things we mentioned. The vibes are really, really high. It's been fun. I yesterday was a really fun day for us cause, like, uh, finally, again, I got, like, a s a whole slew of texts from friends and and people who were just like, Amazing model, thank you. Like, one of my old bosses, um, messaged me, and he was like, Opus 55 just saved our butts in an incident, right?

And it's like, Okay, great, like, we're we're back is really the energy that we feel. And so I think things that, like, maybe you wanted to experiment with something and, like, cost was something that was front of mind for you, right? And, like, token efficiency and these sorts of things, like, that kinda might have held you back from, like, really, like, firing it off and, like, letting a model run. I think that's the those are the sorts of problems you should return to now with Opus 55, cause I think you'll be able to get just, like, a lot further than you maybe otherwise would have been able to with, um, models that were less cost effective.

I think you should try something with, like, a multi-agent setup. I think you will find it to be very good.

A any any other hints, by the way? I I mean, multi is just very broad.

Uh, nope, I think it would be very curious to hear what people build. Um, but yes, that it's, uh, it should be extra good, so

Yeah, including, let's say, uh, roles that are not strictly defined, but they, uh, assign among themselves, right? Which is, uh, to me, the the new thing. Like, you can assign roles, you could always assign roles. Uh, you can give tokens jobs. indeed. Uh, but, uh, this is, like, they're to giving themselves jobs.

I guess, um, say more on multi-agent, right? Like, should we use sub agents? What do you how should people think about multi-agents?

This is pretty good in the sense of, like, basically, generally, you wanna have 1 lead, and you wanna, like, split up, like, a bunch of different stuff. And as long as you can have a task that you think can be generally parallelizable and they share uniform context across it, I think you will get a lot of gains. Now, of course, it depends on the problem that you're pushing at, but, like, increasingly, this is a very superior way for us to handle a lot of these things, and especially if every sub agent is or whatever you wanna call these, like, sub leads, they are asynchronous, um, and they're not blocking of each other.

That is, like, the only 2 bits I would say you should care about in your architecture. Other than that, like, let it rip and, like, see what Claude designs for you. And, increasingly, I think a lot of this architecture stuff will fall into this direction. Like, Claude will get smarter and smarter. It's going to learn how to figure ways to, like, split itself, clone itself, adjust itself to the architectures that make sense. You just need to prevent it you just need to make sure that it can, like, parallel process.

So it needs a shared state, and it needs to, like, not block its sub agents. And as long as you contain those 2 principles, honestly, I think people should experiment and probably try some really crazy stuff. And, uh, I wouldn't be surprised to see some, like, really interesting things on top of that, that maybe they felt like there was some intractable problem that was really annoying and they keep pushing for that, something like this could actually probably just, like, crush in maybe 1 shot or something.

Excellent. Hm.

All right, everyone, go build.

Cool. Cool.