Ep 62: CEO of Cohere Aidan Gomez on Scaling Limits Emerging, AI Use-cases with PMF & Life After Transformers

Aidan joined this week’s Unsupervised Learning for a wide-ranging conversation on model architectures, enterprise adoption, and what’s breaking in the foundation model stack. If you’re building or investing in AI infrastructure, Aidan is worth listening to. He co-authored the original Transformer paper, leads one of the most advanced model labs outside of the hyperscalers, and is now building for real-world enterprise deployment with Cohere’s agent platform, North. Cohere serves thousands of customers across sectors like finance, telco, and healthcare — and they’ve made a name for themselves by staying model-agnostic, privacy-forward, and deeply international (with major bets in Japan and Korea)

Speaker names marked ? are inferred from speaking time and the show’s known hosts — they are not voice-verified.

• 0:00Jacob Effron?

  I'm Jacob Effron and today on Unsupervised Learning, Aiden and I hit a bunch of things, including why he thinks the best application companies will also build their own models. We also talked about the type of data labeling that will matter to make models better going forward. And we hit on why Aiden hopes transformers aren't the final architecture in AI. One ask before we dive in, if you're listening to the show on Spotify, please consider leaving a rating. Ratings help boost the podcast, which in turn helps us get the best guests. Now, without further ado, here's Aiden.

• 0:28Jacob Effron?

  You know, Aiden, thanks so much for coming on the podcast. I really appreciate it.

• 0:31Speaker 01

  Yeah, stoked to be here. Thanks for having me.

• 0:33Jacob Effron?

  Amazing. Well, I figure there's a lot of different places we can start, but one of the most interesting, I think, that everyone's thinking about today is you know, the future of kind of like enterprise AI adoption and usage. And obviously I know this is something you think about a lot. You know, one thing that I feel like I see is there's these different models for enterprise deployments of Gen AI today. It feels like some are just straight up consulting, like you have a big Accenture team come in and build something. Others feel like they're maybe more Palantir-esque where there's some forward deployment engineering, some product being built. And then others are trying to just sell an out of the box product, right? Say, here, here's, so then you go, go make it work. You know, obviously it's a rapidly evolving space, but I'm curious, as you think about which of these models ends up winning out in five, 10 years,

• 1:15Jacob Effron?

  you know, what do you think it looks like for folks working with enterprises in the long term in Gen AI?

• 1:20Speaker 01

  In the long term, I do think that something in the middle is going to win out. It's a new technology. It's very complicated.

• 1:30Speaker 01

  And I wish it weren't the case. I wish it were much easier for enterprises to adopt the technology. But sometimes they need help. So some degree of support is just going to be necessary to integrate it into, into the economy, into all of these different companies, and targeting all the different applications that they might need to go after. I also think AI is unique and these agents are unique in that an agent can only do its job effectively at driving automation if it has access to the same things the humans have access to. And so that demands a lot of access. It needs to be able to see into your emails, you know,

• 2:13Speaker 01

  your chat, your calls, your CRM, your ERP, your HR software. It needs so much context. And so that presents a couple of different problems. The first is privacy. There's very little software that needs that degree of access. And so privacy is a way bigger issue in AI and agents than it is in other types of enterprise software. So that's something that I think we do uniquely well. The second piece is each company uses a different tapestry or like mosaic of software for their humans, right? Like, there's no standard setup of a company of what stack they run on.

• 2:58Speaker 01

  And so each one is different and each one requires some degree of custom setup to bring all of that context together and integrate it into the model. It's another thing we're spending a lot of time on is making our new agent platform, North, much more easy to customize and integrate with whatever that tapestry might be for each particular company.

• 3:24Jacob Effron?

  I mean, obviously, today, there's a lot of like custom integration. You have to build rules around, you know, who has access to what, obviously, you know, a bunch of the VPC deployments as well. In the future, do you have like AI agents that kind of like obviate some of the complexity behind this? And it's like, hey, just enter the tools you want to integrate with, enter kind of some of the basics or like, is that kind of a little bit fantasy?

• 3:44Speaker 01

  No, I mean, definitely like that. That would be extremely helpful if setup could be completely self-serve and the agents just go off and do what's necessary. I think we'll probably have some middle ground. It won't be the extreme of, yeah, there's no humans involved. And actually, we can just install a copy of it and you tell it what you want it to do and it'll install itself. There will be some middle ground where. You know, we'll be able to automate parts of the problem, but not the entire thing.

• 4:18Jacob Effron?

  Obviously, the stakes of a mistake on access to like salary data or something or whatever it is are high enough that you probably want a fair amount of guardrails there.

• 4:26Speaker 01

  Yeah, I think so. Yeah. And definitely customer data, you know, patient data, these sorts of things. You just can't make mistakes.

• 4:34Jacob Effron?

  Well, one thing I think is so interesting about the seat that you have at Cohere is obviously you work with a ton of different enterprises doing different things. It feels like the debate in Silicon Valley these days is what's experimental budget, what's real budget? You have like this great lens into the stuff that's actually working today. How do you kind of categorize or even list the enterprise use cases for GenAI today that you feel like have product market fit?

• 4:54Speaker 01

  It's really hard to categorize. I think it's, there's some vertical applications, like for example, in healthcare, we really want to make note taking and form filling much easier for doctors. And so, you know, having passive listening mics and being able to listen to that doctor-patient interaction and pre-populate things so that they're not spending half their time, you know, typing notes and that type of thing. I think that's very vertical specific, but much more of what we see are kind of these general categories of customer support is something where the technology is ready and the need is very

• 5:30Speaker 01

  much there. And so that's moving quite quickly across verticals. In telco, in healthcare, in financial services, everybody needs this capability and the technology is there for it. The other one, I don't know really how to categorize this, but I guess like research, augmenting humans with an agent that can go off and do a month's worth of research in an hour or two. We're seeing a lot of demand for that sort of capability. So you can imagine if you're in a bank and you have wealth managers, the wealth manager might

• 6:09Speaker 01

  manage a pool of 20, 50 different clients. And those clients might call that wealth manager up and say, hey, I want to hedge against this geopolitical event that might happen in the future. And that wealth manager then needs to go and do research and come up with a strategy of how do I actually even hedge that thing? And it's super time sensitive because that might be days away.

• 6:35Jacob Effron?

  Especially in this current moment, certainly.

• 6:37Speaker 01

  I, yeah, exactly. It's very relevant right now. But if we can make that. Dramatically more effective because these models can read hundreds of thousands of times as much as a human and come back with a very robust piece of research with citations back to all the source documents so that the human can audit it. I think we can just make knowledge work 10x more effective and productive.

• 7:03Jacob Effron?

  I guess one question that folks debate a lot is, obviously, we're so early in exploring the capabilities of this current generation of models, right? We've only had like GPT-4 for a few years, the reasoning models, since the fall. Does it feel like there's a trillion dollars of value just waiting to be unlocked? Have we just froze model capabilities today? Or does it feel like there's actually, we're still going to continue on this curve of model improvement to like really realize the full vision here?

• 7:28Speaker 01

  Reasoning is so obvious. You know, like, it's insane that anyone would be content with like non-reasoning models. Because what is the input space? It's language, it's everything. It's everything from like, hey, what's one plus one, chatbot, to, you know, proof from Aslaz theorem. And so the input space is like, everything. And we shouldn't expect the model to spend the same amount of time to answer those two questions. So the fact that reasoning unlocks a capability for the technology to spend different amounts of energy on different complexities of problem is so obvious and intuitive.

• 8:12Speaker 01

  And we know we have to have that capability built into the technology. There's more that's missing, for sure. For instance, the status quo with models, we spend all this money training them, and we spend $100 million building a model, and then we get our final checkpoint, final weights, and we just distribute that to the world, and they're frozen. And everyone is speaking to the same version of the model, you know, it doesn't remember your chats from months ago. And so there's no notion of learning from experience. And so that's a clear capability that humans have, right? Like, we can start dumb at something new and then become an expert over the course

• 8:56Speaker 01

  of four or five years of practicing that. These models should have the exact same ability to learn from their experience out in the world and learn from feedback from the humans that they interact with. So there are these like... Obvious properties of intelligence that are missing in the technology and that will need to come. But I do think there's a change that's happening right now, which is the whole like scale is all you need hypothesis is breaking. We're very heavy into diminishing returns of capital and compute. And we're just seeing we're going to have to become smarter and more creative in order

• 9:39Speaker 01

  to unlock the next step up in the technology. But I think that's good. It's good pressure on us as innovators to go out and explore and try things out. The old strategy was boring and dumb. So I'm excited for the new era.

• 9:58Jacob Effron?

  It was all just throwing more money at compute, a kind of understood game versus trying to figure out how in the world to make custom compute work for non easily derivaable domains. And obviously some of the algorithm breakthroughs other folks are working on. I think it's seems like an exciting time on the research side right now.

• 10:14Speaker 01

  Definitely, definitely.

• 10:15Jacob Effron?

  I mean, I guess you kind of alluded to it, but I'm curious, like, I think another question folks are asking is, you know, are we on a path to kind of one model to rule them all and you use it, you know, if you're a bank or a healthcare institution or, you know, a world of specialized models for different use cases. You know, obviously, I'm sure you have a strong opinion on this, both on the fine tuning, but also does it ever make sense to have different kind of pre-trained models for these different domains? And how do you kind of think that develops over time?

• 10:38Speaker 01

  I mean, I used to think much more we needed specialized models with MOEs. Models are kind of able to self-develop experts. And these are just like sub-networks within the model. And so that's alleviated some of the pressure. I would say custom models are still important. There's still fundamental context about a particular business or a particular domain that are missing from models that are built off of the web, which the web contains a lot of information about humanity, history, culture, science. But there's stuff that is not on the web. And these models need to be able to get good at that, to provide value in the world.

• 11:26Speaker 01

  So I think that's where custom models make a lot of sense is closing that gap. There's not a lot of manufacturing data out on the web or customer transactions or detailed personal health records. And so for these sorts of modalities or types of data, what Cohere does is we partner with the organizations that do have this data to create a custom model that only they have access to, that gets very, very good at working in those domains. But the general models are extraordinary and synthetic data is able to close the gap considerably.

• 12:05Speaker 01

  And so I think certainly there won't be tens or hundreds of models operating within an organization. You might have a handful. But I don't see it being like every single team is going to have their own fine-tuned model.

• 12:24Jacob Effron?

  Yeah, basically if they focused it on some sort of different type of data that the model haven't been exposed to, you may do some fine-tuning or even some basic pre-training on it, but otherwise, you know, yeah, otherwise it doesn't make sense to have throughout the organization. I guess you kind of alluded to the data side there, and I think I'm curious, obviously in this first wave of post-Jet GPT moments, it was like a bunch of RLHF data. Now it obviously seems like it's moved more toward like expert data labelers in coding, kind of more reasoning tasks. Synthetic data, which you alluded to, has obviously played a big role. Like, what role does data labeling still play for like folks like yourselves that are building model providers and like where are the interesting areas going forward? Is it still relevant in the synthetic data world or?

• 13:04Speaker 01

  Eval. Humans are still like just far and away the gold standard.

• 13:08Jacob Effron?

  Well, I guess definitionally, if you're building models for people, they're probably best suited to evaluate their usefulness.

• 13:14Speaker 01

  Exactly. Yeah, eval is where you just can't take a human out of the loop. Yet. I mean, the way to get the human out of the loop is to have an expert that is better than your current model observing it. But that assumes you can build that expert in the first place, so why do you? So there's, I think, a hard dependency on humans within eval. On the data gen side. It's too expensive. Yes, you're right. Like, we definitely still need human data. But it's too expensive to go find 100,000 doctors and have them teach the model to do medicine. That is not a viable strategy.

• 13:58Speaker 01

  It was a viable strategy to teach the model general things about like how to converse, chit chat, that type of thing, to find 100,000, you know, average people and they can teach the model to do that stuff. So we've had to become much more creative. But teaching the model to chit chat and do all that stuff unlocked a certain degree of freedom in terms of synthetic data generation, which we can then apply to these specific domains like medicine. And we can use a much smaller pool of human data. You know, maybe I go to 100 doctors and get them to provide me some lessons and teach my model. And then

• 14:38Speaker 01

  I use that pool of like known, good, very trustworthy data to generate a thousand-fold synthetic look-alike data. And like you say, in verifiable domains like code and math, it's way easier because you can you can check the results and use that to filter the synthetic data and pull out the garbage and, you know, find the gold. In other domains, it becomes much more complex, but it's it's still viable. So at this stage, I would say in terms of like the overall data that Cohere is generating for a new model, an overwhelming majority is synthetic.

• 15:20Jacob Effron?

  Super interesting. And you kind of alluded to, obviously, if one of the questions that I feel right now is, you know, how far test time compute gets us and, you know, what spaces it will and will work in. Like, how do you kind of conceptualize that today? Both, you know, what works well today and then also, you know, on the on the near term frontier?

• 15:34Speaker 01

  For test time compute, for reasoning, like Cohere cares way less about solving Math Olympiad. Like, we want these models-

• 15:43Jacob Effron?

  You don't have enterprises coming to you begging for for models that can do well in Math Olympiad?

• 15:47Speaker 01

  No.

• 15:50Speaker 01

  But we do have them begging us to help them automate processes within the back office that use this piece of software and also go out on the web and do some research and then do this. And so for us, it's very much about teaching these models to reason through solving the problems that exist within business using the tools that humans in business has used to get them done today. That's what we care about. And it has been a complete step change in terms of improvement. There were tasks pre-reasoning. We just couldn't get a model accurate enough to accomplish that. It was like impossible. Models could not solve the problem.

• 16:34Speaker 01

  They almost always fail. And with reasoning, they almost never fail. Like, pretty much everything you throw at the thing, it'll figure out a way to actually accomplish it. So reasoning's ability to unlock reflection and understanding why the first attempt failed and then using that to find another path, an alternative path to the same outcome, that's been a real unlock.

• 16:59Jacob Effron?

  Yeah. Was there like some sort of crazy moment for you when you were initially playing around with these reasoning models and saw that on one of these impossible tasks?

• 17:05Speaker 01

  Yeah, like things that just don't work start working. And complex tasks, there's many delightful moments where you're like, how did it know to look there? You know? And it's, oh, it's because it failed looking in the obvious places.

• 17:20Jacob Effron?

  And then you read the reasoning traces, you're like, ah, it's actually pretty smart.

• 17:23Speaker 01

  Yeah. I mean, the reasoning traces are just, it's like jaw dropping. It's so beautiful that these models organically problem solve and like think through stuff and, you know, they have their own little epiphanies and they're like, oh, you know what? I should check over here and maybe this happened and that's why it's not over there.

• 17:46Speaker 01

  Has been super rewarding to see.

• 17:49Jacob Effron?

  I guess when you're talking to enterprises, how do you kind of contextualize where Coher fits the ecosystem today? Obviously, you've got the really large AGI labs and OpenAI Anthropic. You've got a bunch of open source model providers that are out there and folks working with them. How do you kind of contextualize that? And also, it seems like you build a lot of these, you help these enterprises build a lot of use cases. How far into like the application layer do you want to go? Like, do you want to have like a support product? Or I'd love to hear your riff on that.

• 18:18Speaker 01

  Yeah, so with North, we're pushing into the application layer. And that was totally motivated by, we were selling models and we just kept seeing our customers building the same thing again and again and again. And it would take them like a year to build an application that took models, presented a UI, integrated with all the different tools and data sources, did the privacy and ACL inheritance stuff well. And then it was usually built by the AI team inside of these enterprises. And so they're not product teams. And so it's actually not that nice of a user experience. And so the people within the company wouldn't want to use it. And so we said, what if we just solve that entire problem for the enterprise?

• 19:02Speaker 01

  We're going to create a like truly consumer grade product experience that people love to use, is super intuitive, extremely low latency, but ships with all the features that an enterprise needs, like the ability to customize it completely. You can customize the UI, you can rebrand it however you want, you can customize the data connections and the sources that your model is able to pull from and the tools that it's able to use. You can even, if you want to, like, let's say you've trained some llama fine tune or something on your data, you can plug llama in there and you can expose it through this application. And so it's really just setting a company forward 12 to 18 months in its

• 19:46Speaker 01

  product roadmap. And they immediately can distribute this tech into the hands of their employees. So that's been the push for us. Definitely.

• 19:57Speaker 01

  Cohere has some very nice strategic attributes like the fact that we're not locked into one ecosystem. We're not within one hyperscalers ecosystem. We're able to deploy anywhere. We also release our weights non commercially. And so we're happy to deploy in VPC because we're not worried if you can access our weights. We're very happy for you to do that. So I think just far and away, we are the best partner for enterprise.

• 20:24Jacob Effron?

  I guess I'm curious because obviously one of the cool things about Cohere is that both you have like the model development side and cutting edge, you know, you have great research team and then you also have kind of the applications that you build on top of the models. You know, obviously there are some folks that would say, hey, like we'll just take the best models that are out there and we'll build the support application or we'll build, you know, whatever it is and we don't do anything on the model side. We're just like, we'll take what's in, we'll do some like fine tuning on an open source model. How do you kind of articulate the advantage you get from building both the models like and the applications versus these folks that are just kind of building applications on top of the underlying models?

• 20:55Speaker 01

  I think we have far more levers to pull to deliver the experience that the customer needs. Because we're vertically integrated, the next version of command, our generative model, is going to be optimized, actually the latest version already is optimized for the use cases that we know our customers need in North. So it knows how to use an ERP, it knows how to build, it knows how to build with a CRM. And so that integration between the technology, the models, and then what the actual customer needs is something that I think is crucial to product quality.

• 21:35Speaker 01

  And it's part of the reason why a lot of the existing applications are, in my opinion, missing or not fulfilling the product promise that they're making to their users. They're a consumer of the technology. They can grab a, you know, some Llama weights or something, but they're not able to change that technology for the needs of their customer. And so that presents a Fundamental barrier to product quality.

• 22:03Jacob Effron?

  I'm curious, like, one question I feel like gets debated a lot in the venture world is like the right teams to build these AI applications and like how much AI expertise you need. Obviously, like, you could imagine a spectrum of just like smart Stanford grads that, you know, know basically how to fine tune these models but are not super deep on the research side, all the way up maybe to a transformer paper co-author like yourself. You know, do you think generally, as you think about like the right teams to build these AI applications, like, across the board, do you think you need this kind of deep level of model knowledge or how do you kind of think about that?

• 22:35Speaker 01

  I mean, if you look at the ones that are building the best products, they do have that knowledge.

• 22:41Speaker 01

  They do. There's just like a deep familiarity with how to build language models. And even if they're not training models by themselves, they're trying to find ways to like approximate that and have as much of an impact at the model layer as they can.

• 22:56Speaker 01

  So I definitely think that's a crucial piece to a company building on these models succeeding. We try to play that role for our partners. So we power a bunch of AI features within the Fusion Application Suite with Oracle and a bunch of other large enterprise SaaS companies. And so what we can do is because we can intervene at that level, whatever they need, we're going to make it happen and we're going to make it work for them.

• 23:25Jacob Effron?

  What use cases does it feel like are next? Obviously, you mentioned the ones that are working today. I'm curious, you know, do you think about the next 12 months, both with the current set of model capabilities, but also where it seems the trajectory is going? Like, what do you think? We're having this conversation in a year and I asked the same question of what has product market fit? Like, any that you're like, yeah, these seem pretty close?

• 23:42Speaker 01

  I think any of the like deep research style use cases, the tech is ready and the market is just being blown away by it. It's, they can't believe it. It's surreal.

• 23:56Jacob Effron?

  Yeah.

• 23:57Speaker 01

  Like, this is something, the model comes back with a report that would have taken a month and a half from someone who costs a lot of money and it comes back in 30 minutes, an hour. So I think that very much is like, it's ready for prime time. This is going to be integrated into every single enterprise in the planet. It's something that we're pushing super hard. Beyond that, I think looking forward.

• 24:22Speaker 01

  I actually see more of the mundane back office stuff starting to come online. The easier it gets for an enterprise to automate this technology, when the infrastructure for doing that automation gets in place, like North, so a system that has access to all the different tools and pieces of context that it might need access to, as well as a user experience to build those automations, to ask of the model, some task, accomplish some task for me. As that infrastructure gets installed and plugged in at these enterprises, I think we'll start to see a lot of the back office tasks start to go, start to get unlocked.

• 25:05Speaker 01

  So that's stuff in finance, in legal, certainly, sales is very much ripe for this sort of thing. Yeah, like if you think about like what a seller does before they go into a meeting with a customer, it's research.

• 25:21Jacob Effron?

  Exactly.

• 25:22Speaker 01

  You're going through like, what is this company, what's their strategic imperatives, who's leading the efforts that I care about selling into? And then you're also doing research internally at the company, what previous conversations have we had with them? How did that go? What were the sticking, you know, factors? And you need a, like an intelligence briefing on the person you're about to meet, everything they care about, everything they've heard about, you know, heard about your company. And that makes you dramatically more effective at your job. Yeah, so I think sales is going to be another big one.

• 25:59Jacob Effron?

  And I guess, you know, obviously you take support, sales, there's a whole host of companies that like, that's all they're focused on, right? Whether it's like, you know, Sierra and Spore, or, you know, Clay and folks on the sales side, like, you know, I assume on the one hand, enterprises would love to work with one vendor that can kind of help them across all of these. On the other hand, you know, if all a team's thinking about all day is one of these problem areas, maybe they build better workflow, like, you know, do you think we, you know, do you think we kind of federate into, you know, hey, I'm an enterprise and I choose best-in-class solution for finance, for sales, for support, or like, do I go to kind of one place that has pretty good general knowledge across all of those things?

• 26:33Speaker 01

  I think there will be like a scattershot phase that then consolidates. So I think initially, it will be federated into all different teams are buying their own little applications, and then they're going to realize.

• 26:48Speaker 01

  Wow, okay, like, you know, sellers don't just exist within their CRM. So I have to integrate that agent into all the other pieces of information within the company. Same thing with the finance team. They don't just exist within their ERP. So I have to integrate. So now you have like this insane maintenance burden of all of these data source connections into all of these disparate apps. And so there's going to be a strong push towards consolidation. I want one platform plugged into everything that I can accomplish all the different automation objectives that I have on top of. So the long game is building that platform, which is what we're doing with North.

• 27:26Jacob Effron?

  Yeah, it makes sense. And obviously, all these things need to speak to each other to like really be able to ultimately answer the questions you want to answer on the research side or kind of like enterprise inside side. You know, I guess reflecting back on the on the co-chair journey so far, you know, what sticks out as like some of the key decision points that, you know, that you had to make that really tipped things that maybe were like, you know, 51, 49 decisions along along the way.

• 27:47Speaker 01

  It's been a lot of consequential strategic decisions, everything from like starting up in Toronto. That's one that has really paid dividends. And it was not a given. We really did not know if it was the right idea early on. We had some arguments for why it was, you know, like Jeff Hinton is in Toronto. The Canadian AI ecosystem is, you know, really has come out of it and like so many of us have. Starting up there has given access to really the world's best talent.

• 28:26Speaker 01

  And it's made us a darling of the Canadian AI community or the Canadian technology community.

• 28:36Jacob Effron?

  I'm curious, you know, to hear you riff on this a bit. It feels like, you know, obviously, there's this really interesting interplay between, you know, you guys have obviously gotten a ton of support from, you know, the global ecosystem in Canada, the, you know, the government obviously seems super set in about what you're building. And I think there's like this larger question about like, it does feel like a lot of countries want to have their like own foundation model companies. I'm curious to be able to just riff on that like intersection of kind of national politics and priorities and foundation models. And do we end up with like, you know, a couple dozen because, you know, each country kind of wants to have their homegrown, you know, player?

• 29:08Speaker 01

  We probably end up with a few. I think the like buy versus build question or like partner versus build question is a hotly debated one. For Cohere, we're very international. You know, my mom is British, my dad is Spanish. I have passports in both of those places, also in Canada. The leadership team is from all over, you know. So we try to be a partner to these countries in adopting the technology and making sure that it works for them. Our partnerships with companies like Fujitsu in Japan, we're deeply, deeply investing in Japanese in our models and we will forever now.

• 29:52Speaker 01

  We have a deep commitment to Japan and making sure that our tech works as well in Japanese as it does in English. We recently announced a partnership with LG in Korea and we're doing the same thing for Korea. So we're certainly in favor of supporting everyone to make sure that their economy can adopt this technology, especially in places where the majority aren't native English speakers or the data that you're dealing with in that jurisdiction isn't in English. A lot of our work, like on the open source side with Cohere 4AI, the Ia project, that was like the largest data collection effort for any machine learning project.

• 30:32Speaker 01

  Thousands of people, native speakers of over 100 different languages, contributed data and we open sourced it so that not just Cohere's model gets better in those languages, but so does every other language model. Yeah, we have a belief that the technology will not be useful and huge swathes of the global population will miss out if it doesn't speak their language and it doesn't understand their culture. So we've invested very heavily and will continue to.

• 31:04Jacob Effron?

  Super, it's super powerful. I mean, you have alluded to it, but I'm curious, the, you know, the kind of like state of the art, you know, AGI labs, how many do you think we'll have in like five years and what actually ends up making one better than than another?

• 31:19Speaker 01

  We're all starting to do different things. So there's this like.

• 31:27Speaker 01

  Differentiation or like the diversity of strategies emerging. OpenAI is right now clearly pushing the consumer front. You know, Gemini is coming for that crown. I know that Metas is planning the exact same thing. Anthropic is very good at code. Like, that's clearly the thing that they are the best at. And we are focused on enterprise and sort of business back office applications. So everyone's finding their lane and their niche. I don't know. I think among foundation model companies, there will probably be a couple handfuls.

• 32:12Speaker 01

  And that will cover a lot. And then we'll see new generations of foundation models come out, right? Like, foundation models for biology, for chemistry, for material science, for... We'll start to see new generations of the stuff. Yeah, I sit on a advisory board for Kaya. It's a cancer data and compute sharing alliance with like the Fred Hutch and a bunch of other large American cancer research and treatment centers. And that is data that is not out on the web.

• 32:49Jacob Effron?

  Yeah.

• 32:49Speaker 01

  Right? And there's this like question of, okay, what if we had like a GPT moment where massive capital flooded in that was applied on top of that data or those modalities that count for treating cancer? And how do we build conviction about that? How do we de-risk that in the offset? If we do need to deploy $10 or $20 billion of capital into this thing and we could solve cancer, you know, let's say that that's the outcome that we're driving towards. What are the steps along that that we can show? So I can already see... Those ideas emerging.

• 33:31Speaker 01

  And as LLMs and these sort of like general image, video, audio, language models emerge, there are other quite different domains, like protein sequence modeling, material science, et cetera, that need a similar global effort to yield the same sort of just incredible advancements.

• 33:56Jacob Effron?

  Yeah, and it seems that the big thing there is ultimately like you might need data generation, right? Like you don't have tokens on the internet. And so even, I mean, obviously, a lot of problems with cancer data is that it's not particularly well structured or linked or tied to outcomes. But like also, you know, for a lot of these bio foundation model companies, you know, they all spin up labs, right, to actually generate a bunch more data. Or, you know, the robotics side, it feels like there's a similar issue to you. And so I think it'll be, you know, I think maybe that ends up needing to be the first area of investment. And then, you know, also, you obviously don't get the benefit of getting to evaluate right away. Sometimes it takes, it takes five, 10 years to figure out if it actually worked. But seriously, a fascinating set of problems that I hope the best and brightest are going after.

• 34:36Speaker 01

  Yeah, yeah, no, I, me too. There is a lot of data out there in these domains. I don't know if there's like a token scarcity. It's just that it's siloed and locked up in 100 different places that refuse to share with each other or talk to each other. The data absolutely exists in sufficient quantity.

• 34:57Jacob Effron?

  Interesting. So a human problem more so than a data generation problem?

• 35:01Speaker 01

  Yeah, exactly.

• 35:02Jacob Effron?

  Sounds about right. I guess I'm curious also what you think of some of these like newer foundation model efforts. Obviously, you know, we have SSI and thinking machines, like, you know, felt like for a while we might be done with like net new LLM providers. And then obviously, there's been a few in the last year. Like, do you think we'll see more of that? Any reaction there?

• 35:19Speaker 01

  Yeah, I mean, I hope that tons of new companies come out with new takes on stuff and try their hand at building something valuable for the world. I think that's great.

• 35:32Speaker 01

  On like the technological front, there's a lot of debate about the architectures that could come next and like are better architectures out there.

• 35:41Jacob Effron?

  Think you'd have a good take given you and that did the current one?

• 35:44Speaker 01

  Listen, I think I've said this publicly before. I'm like the first person to be like, why the hell is the transformer still being used? What's going on? Yeah, I I've been waiting for it. Like we named in our New York office, we have a a meeting room called SSM. Because I was like, this is it. Transformer is dead, you know? Let's go SSM. But then it turned out you can just kind of like steal the good bits of an SSM and put them into a transformer and then the need to swap to an SSM goes away. Now there's these discrete diffusion models that are coming out and they're a super cool UX. Like you start, it's like normal diffusion, right? Where you have a wall of noise, like a wall of noisy tokens and text.

• 36:27Speaker 01

  And then out of the ether emerges the response. And I mean, that's cool. Is it actually a better language model than a transformer? I don't know. I don't see why that would be the case. So it's tough, man.

• 36:46Jacob Effron?

  How likely do you think it is that we do have a new architecture in like the next five, 10 years it becomes predominant?

• 36:52Speaker 01

  Oh, God, please, please. You know, like, if you had have asked me in 2018, a year after the transformer paper got published, what was the likelihood that in seven years we'd still be using transformers? I'd put it pretty close to zero. So I'm not going to make an estimate. But yeah, the longevity of this thing has really surprised me.

• 37:24Jacob Effron?

  You've been doing, obviously, cutting edge AI research for a long time now. I'm curious, what's one thing you've like changed your mind on in the last year?

• 37:31Speaker 01

  Maybe the the scaling hypothesis.

• 37:35Speaker 01

  I feel like I was like pretty loyal to that. The past few years.

• 37:42Jacob Effron?

  Good reason to be loyal to that, definitely.

• 37:44Speaker 01

  Maybe that's something I changed my mind. I mean, in the back of my head, I've always thought there's like, really? Can it like, are all the capabilities we need going to fall out of scaling? Really? Like, it seems unlikely. But the evidence just kept showing up. It was just, yeah, we made it bigger. Now it can do math, you know, like all the sorts of sorts of stuff. So I was like, yeah, I guess this is it. But now we've just been shown, really beaten over the head with it, that scaling is not going to get us there.

• 38:18Jacob Effron?

  Yeah, but some people would say all the capabilities will fall out from scaling test on compute, right? That it's just like another scaling vector. Like, do you subscribe to that? Or is there still some algorithmic breakthrough needed on the on the horizon?

• 38:27Speaker 01

  I mean, then you're like really muddying the definition of scaling. Because a lot of like what scaling is now is actually data and getting diversity in the data, finding demonstrations for the model of how it can problem solve in specific domains. I don't know if it's as trivial as the previous scaling law, which was literally like, I'm going to build a 2X larger computer and press run on the same thing and the loss will go down.

• 38:55Jacob Effron?

  Does anything change on the on the hardware side now as you think about needs for, you know, for hardware going forward, given this kind of, there were maybe out of the pre-training era and into, you know, more test on compute and other approaches?

• 39:06Speaker 01

  Yeah, test on compute still requires tons of compute. It makes inference, whatever it is, three to 10 times as expensive. And for training, it still requires just tons and tons of of compute.

• 39:26Speaker 01

  But I think compute is going to get cheaper and more abundant per flop. There's multiple options now for training compute, which there wasn't before. And so that's a very positive thing for the world and for the industry. We can wire together more than just one type of chip and get a very effective supercomputer for building models with. So all the trends on the compute layer of the stack are very good for those of us who have to consume a lot of it.

• 40:01Jacob Effron?

  What are the future model milestones that are meaningful to you? When the latest state of the art model comes out, is there something you try? Or like, I thought the playing Pokemon was a funny valve from Claude. Not that any of your enterprises particularly care about that. But like, anything that you try whenever one of these new models comes out, just to get a, that you think will be a meaningful capability breakthrough?

• 40:21Speaker 01

  I mean, for my own models, like for command, the first thing I do is just, I try to use it as part of my day to day work. I try to get to automate stuff that I have to do.

• 40:32Jacob Effron?

  What's the latest thing you've automated?

• 40:33Speaker 01

  I do a lot of prep for meetings using the model. And basically, like, we have an internal, in North, like, deep research, but it's plugged into not just the web, but everything. Like, it can see every single call transcript, everything that I've said, every conversation that I've had in DMs, all over the place. It just sees as much of my life as I can possibly make it see. And it's able to give extremely compelling responses. So yeah, that's most of my day, right? As a CEO, like, I'm meeting customers, investors, talent, and I need to be ready for those meetings.

• 41:17Speaker 01

  And so that's where I plug it in. I can't do that with other models. So usually when I'm playing with like, A new third-party model. I do like riddles and stuff. I don't know. So the reasoning, the reasoning models were very good.

• 41:36Jacob Effron?

  Do you have a Go to riddle?

• 41:37Speaker 01

  I do. It's one of those like, you have to come up with a matrix and like, figure out who is what, based on all this information and, you know, who checks what boxes. And it's about spelling words. And so the ability to break a word down into its characters, similar to like the strawberry thing, which we know models suck at, is involved. I usually go to that one. And then math stuff, just like, can it solve math problems? But it'll shift, right? Like, as models become able to do all of that, like they are now, I've stopped using the math ones, because it solves all the math ones. It's not interesting. Like, there's nothing to ask there. Yeah, most of it is about trying to break it, right?

• 42:21Speaker 01

  Like, you're trying to find the boundary of the application of this model.

• 42:27Jacob Effron?

  Well, I mean, shifting gears, you know, I feel like, obviously, you were part of the, you know, original Transformer Paper crew, and, you know, have been a part of a ton of really interesting AI research work. I feel like people like to write these thought pieces about, you know, the culture that enables these things. I'm sure you think about this a lot in the comments, to building the Coher research team. What have you learned, I guess, or what lesson do you take away from, like, the most successful AI research groups you've been a part of?

• 42:50Speaker 01

  With Google Brain, what was so incredible was it was a home for really brilliant people to do their best work. You got pretty much complete research freedom, to do whatever you wanted to do, tons of compute, tons of software infrastructure to use that compute, and really great people to do it with. And so I think that setup led to some really interesting things. But it's not the setup that delivers an incredible product.

• 43:30Speaker 01

  And so for Coher, the needs are different. We have a very targeted focus. Like, there is something to accomplish. We have an objective, which is driving this automation, making this technology good at using all of this different software. And so the problem set is quite focused. You can't work on whatever you want. There isn't time for that. And so we have to be much more focused and targeted in terms of where we invest. But what we preserve are incredible people, tons of compute, and the same sort of world-changing ambition that I felt back at Google. And also, a warm culture.

• 44:13Speaker 01

  I think maybe things have changed in different labs. But when I was there, it was like a deeply warm and meaningful, welcoming place to be that cared about something. And so that's something that we, I think, we preserve quite well.

• 44:38Jacob Effron?

  Yeah. No, that's awesome. You know, is there an impact of, like, continued model improvement that you think we're underthinking right now as a society?

• 44:46Speaker 01

  Learning from experience. Once models can learn from interacting with the user, just from like a, like what you can build perspective, that capability unlocks so much. Right now, with a model that doesn't do that, I give feedback, but when I start a new chat, it's forgotten all of that feedback. It's just wasted time. Like, if it can't do something, I'm just annoyed, and then I close the window and I go do it myself. But if a model is kind of like an intern, where, hey, it's the first time it's doing this thing, it's gonna mess up a little bit, I'll teach it, I'll guide it, I'll give it some feedback, and then, hey, look, you did it, and it never makes those mistakes again.

• 45:28Jacob Effron?

  How does that happen? Is that like, you have your own personalized model, or is it just stuffing that in the context window, or like, putting it, you know, in some database that you tie, you know, via RAG or something, or how does that actually get actuated?

• 45:39Speaker 01

  Well, it doesn't exist yet, so I think people are experimenting with all of the above. I think it probably looks like the Was it the last one that you said, which is like putting it in a database that can be queried and looked up? And there's just always a context of the previous history of interactions available to the model when it's generating. Probably looks something like that. But yeah, you think about what that unlocks, like I'm so much more invested in interacting with this model because it's learning from me. It's growing with me. It starts to know me, know my preferences, know how to do the things that I need it to do because I've spent so much time teaching this thing. And I've turned it from my little intern to like me 2.0.

• 46:22Speaker 01

  And so that connection with the system is something I'm, I'm very excited for.

• 46:28Jacob Effron?

  Right now, you're right. It feels like I've come into the model, a new batch of interns that just came in in their first week. And then you've like do it for a week and then it's again, like someone else's first week. It's like roundhog day to some extent. You know, obviously, I'm sure in the course of your work with, with models and general AI research, you kind of probably come to believe that the advancements in AI are going to happen sooner than you previously expected. Has it changed, has that changed at all, like the way you live your life or like think about, you know, the, the future?

• 46:56Speaker 01

  Not remotely.

• 46:58Jacob Effron?

  Yeah.

• 46:59Speaker 01

  Not at all. Not at all. I'm super excited. I'm super excited. I think, you know, my dad is a cancer survivor, you know? I really hope, you know, he lives for, for a lot longer and that treatments continue to get dramatically better. I hope that, I hope that we're able to drive costs down and that supply increases in the world. Cause I think we are supply constrained. We're not demand constrained, which is why I don't think there's going to be any sort of like mass unemployment. I really don't believe in that.

• 47:44Speaker 01

  So long as we're able to move people onto the spaces that we need them and we can retrain folks, humanity has infinite demand. And so we are fully supply side constrained. And so this technology isn't going to displace people. It's going to augment them and let them do more and deliver more of what the world is wanting and needing. So I'm, I'm really excited about the future, but I don't think it's a.

• 48:13Speaker 01

  You know, utopia.

• 48:17Speaker 01

  I think it's a much better world. It's a much better world, but it's not a utopia. It's not like you should be, you know, liquidating all your assets and just going on vacation permanently or something like that.

• 48:29Speaker 01

  Yeah, no, it's just going to be a way better world.

• 48:31Jacob Effron?

  Yeah. To what extent do you worry about like some of the AI risks that folks talk about, both, you know, in the short term and then like, you know, the existential ones?

• 48:38Speaker 01

  I worry about them.

• 48:42Speaker 01

  I worry what sort of capabilities bad actors will be able to access, especially at the state level. I want to make sure that liberal democracies are the first ones to get it and that they establish an advantage.

• 49:00Speaker 01

  I worry about, do we have the infrastructure in place to facilitate moving people to new careers if particular jobs are impacted? I want to make sure that we have that, that capacity to move them and retrain them and, you know, put people on new, more fulfilling

• 49:23Speaker 01

  work.

• 49:26Speaker 01

  But I'm not afraid of like the X risk, like Terminator takes, it breaks out of the box and gets all of our nukes and blah, blah, blah. Or it's like nefariously manipulating people to do what they want. I, I really think there's so much to worry about with this technology that the doomsday scenarios aren't where we should have the public and policymakers focus right now.

• 49:54Speaker 01

  There's more than enough we've got to get right in the near and midterm.

• 49:58Jacob Effron?

  It's been a fascinating conversation. I'm sure there's a ton of threads folks will want to pull on. I'd love to just leave a last word to you. Like where can folks go to learn more about you, the exciting work you're doing at CoHere? Anything you'd like to leave our listeners with? The, the mic is yours.

• 50:10Speaker 01

  Yeah, I'm not that interesting. I think Coher is the interesting one. And so you can go to coher.com. You can follow me on Twitter and you'll get all the alpha on AI. AI stuff. Love it. Thanks for having me, Jake.