Bret Taylor on AI and the Future of Software | Ep. 42

[00:00] Clearly in three years, we could talk about what are the best practices to set up a software team that's optimized for this technology. We'll know what those best practices are. And right now we're just figuring them out in real time. And like my hypothesis is the companies that figure it out first will move the fastest. It's fascinating to me. Brett, thanks a bunch for doing this with me. I'm super excited for it. Thanks for having me. [00:30] Apocalypse, if we can call it that. Sasmageddon. Sasmageddon. So basically, it's like, you know, in public markets, all of these companies are trading way down. You know, you go on X and everybody's talking about how like, you know, software can now be written in two seconds. And so there's no moats anymore in software. And so it's leading a lot of people to ask, like, where does durability come from? And so I just wanted to sort of start with this topic because, you know, you've built your own companies. You've been the co-CEO at Salesforce. [01:00] AI startups there is, you're on the board of OpenAI. How do you see software like in this moment in February 26? So first, I think the market isn't necessarily reflecting an indictment of individual companies. I think it's more of a [01:15] uh broad view of like the the bigger questions you were saying i.e every software stock is down but i don't think that means every software company is equally disadvantaged it's just basically anxiety about the future i think it's a few things um we can talk about sort of defensibility broadly i think it's a really interesting question i think if you look at the history of enterprise software a lot of the value has gone to the big systems of record so

[01:45] famously powered in the early days of software. And then you end up with all the software as a service companies, SAP, Workday, Salesforce, ServiceNow. If you look at what a system of record is, it's essentially a database with a bunch of workflows around it. And to date, those workflows are manipulated by people clicking on buttons in a web browser, filling out forms. If you had to like synthesize pre-AI, like why were those businesses so good? Was it the source of truth thing and that there had to be some immutable thing? And so the database row, [02:15] system integrations, like what what do you attribute the success of systems of record to? So I think the reason why a system of record has always been the most valuable is it is the anchor tenant of your technology deployments. You know, if you wanted to, you know, create a workflow for quote to cash or something like that, you had to integrate with your ERP system and your CRM system. So as a consequence, you know, the companies that sort of owned those databases could either [02:45] third-party company, they would often be a part of the ecosystem like Salesforce's AppExchange or whatever the marketplace equivalent is for SAP. And so you ended up with a lot of value in those systems, which meant switching costs were just really high because it was sort of this, that system plus all the partners that integrated with it sort of created gravity and high switching costs. And then similarly, you just end up accruing a lot of value either by collecting rent from your

[03:15] So it sort of became the sun and the solar system, you know, for each of the different lines of business that these systems of record were sold into. And then you'd end up where you'd get a scale. So you'd get sales capacity scale, you know, so the larger you grow, the more salespeople you have. You can reach more and more people. Then there's the proverb, no one gets fired for buying IBM, which, you know, obviously a somewhat dated expression. [03:45] chose SAP. If you choose something new and it doesn't work perfectly, big trouble. Then you're the same. So all those things sort of accrue. But then the question is now that all of a sudden that a lot of those start getting chipped away with AI agents. First, could you just vibe code it in a weekend? So does it change build versus buy? So that's one risk. Does it change when you come up on that renewal? Are you going to make a different decision? Secondly, I actually think the more fundamental thing is what is the role of that system of record if AI agents are doing most of [04:15] around on an ERP system to onboard a vendor, if you just delegate to an AI agent to do it, [04:21] all of that is sort of invisible to you. And all of a sudden it goes from being an application to sort of a database. Right. Similarly, if you imagine a CRM system and rather than having people staring at it all day to manage their leads, contacts and opportunities, if you just say, hey, generate me some leads. In other words, like does a does a system of record have a place in the world if nobody logs into it? And it does. But the real question is like, how valuable is it? How important is it? You know, when you go back to my metaphor on the solar system here,

[04:51] the agents running around it. And it's just really interesting because if you imagine you're running a sales team, how much do you value the database of leads versus the agent that generates the leads? And along, like ancient history, three years ago, those were the same thing. But now you're like, gosh, I actually probably care more about the lead generation and how it's stored and tracked is actually maybe a more tactical part of it. So there's all sorts, and that's true of every system of record. This isn't, I just know CRM systems pretty well. And if you look at ITSM, [05:21] which is like the place where it's now plays or ERP systems, which is Workday, SAP, Oracle, etc. There are all these questions start coming up. And so what's interesting, though, is I think every single one of those companies could transform and benefit from AI. I really do believe that. You know, you saw what Microsoft did in the cloud transformation, and they went from being dependent on Windows revenue to going to Active Directory and Azure and all those other things. But it was really awkward. You know, I think folks like you and me back in the day used to probably dismiss Microsoft. I mean, I certainly did. [05:51] I didn't foresee them becoming as powerful and strong as they are today, but it was good leadership, good technology. But I don't think the market knows who is Siebel Systems and who is Microsoft in this landscape of software companies. Probably no one knows what Siebel Systems was. That was the company that Salesforce beat to become the cloud CRM. So can you actually develop this ecosystem of agents around your platform? And will it become more valuable than the platform you had?

[06:21] zero. I don't necessarily believe that. But you look at all of that, if you're just an investor in public markets, you're like, I'm going to sit on the sidelines. I'm going to let the market play out a little bit. And I think that's sort of what's going on. Totally. Yeah. I mean, you can never know for sure who's going to turn into the next Microsoft, but you can kind of try to think about like who has the structural ability to expand, like who's got the right with customers to make the expansions and then which products will be easier. So like, you know, in the database question, [06:51] Or is it easier for a modern agent to go say, well, I'm going to go build a database at some point because I can I could do that. And I've got the customer relationship. And how do you think about like what creates the rights to expand? I think all the incumbents sort of have a right to win in a lot of ways. You know, in the same way we talked about, you know, why a system of record is powerful. I think you could say the same logic for all the agents right on top. The dynamic that plays out, though, not just with AI is when a new technology comes out like the web browser or the smartphone. [07:21] rarely is the expertise on how to do exceptional things with that technology at the incumbents. So first, there's this thing in enterprise software, there's a phrase called best of breed and best of platform. Best of platform means, hey, we're a Microsoft shop. We just buy Microsoft stuff. And it sounds silly, but actually there's a lot of logic to it. Like, A, you get sort of good procurement leverage. B, everything works together. You don't have to deal with a ton of people. There's probably some benefits, all sorts of things.

[07:51] technologies come out as the pendulum swing from best of platform, best of back to breed, because when the new web browser came out. It's much easier to get a 10x experience. 100%. And also just think of the pre and post web browser enterprise software. You're running client server Windows software. And it's a completely different skill set to make a web application, as you and I know. And so at the time, there's this window of time where best of breed competitors are light years ahead of the incumbents. And it's a race. [08:21] can the best of breed upstarts turn into GitScale before the incumbents figure out the technology? And that's what was going on right now. So I would argue very few of the incumbents have any credible, decent AI technology, but they will. It's inevitable they will. You don't understand. Why is that? What's the real reason for it? Because I see these companies that have... [08:44] let's call infinite resources, roughly speaking. They ought to be able to hire who they want. They ought to know what [08:50] the products could look like. They ought to be able to try them. They are like, why is it so hard for like, let's say legacy companies to like catch up quickly, you know, versus like an AI startup with 50 engineers seems to, you know, outperform, you know, the teams that are 10 or whatever times bigger. But a big company is cultural. Is it systems? Is it? I like the phrase strategy. I don't remember who to attribute that to. We could pull up chat to you. I think the idea is like in these moments of big platform shifts. [09:20] what were your strengths can become weaknesses. So let's just take Siebel systems in the birth of the web browser. They have a, you know, on-premises CRM system. When you say, okay, like let's compete with this cloud native CRM system in Salesforce, you start to say, well, I don't want to start from scratch. Like we've got all these assets. So how do we do it in a way that takes advantage of all of our assets? And so all of a sudden you're like, okay, let's not just

[09:50] Let's transition from this product to that product. And what if someone wants on premises too? And that's our strength. We should play to that strength. And you start basically making all these decisions that sound really clever because you're playing to your strengths. And in practice, if the technology wave is bigger than the category, which I think the web was, as an example, you end up basically chipping away at doing a pure play value proposition. It can also happen with business models, though. [10:20] time you'd have perpetual license software and moving to software as a service that's a huge change for a business to make for your customers it goes from being capex to opex for you as a company it changes ratable revenue i mean adobe uh shantanu did this adobe very few companies could make that transition yeah [10:39] and you have to sell it differently if to compensate salespeople differently revenue recognition is different so you have the product strategy tax you have the business model strategy tax you have even like incentives of sales people there's a strategy tax because you know you don't want to just have your business collapse overnight so you can't just it's so easy for clever or silicon value like just pivot i'm like yeah if you're a public company you have to you know go in front of your investors every single quarter and be like yeah hey guys i know our revenue just [11:09] to turn around next quarter like there's you don't survive that so you just compound all those things and all of a sudden you're like why does a 50 [11:16] Person coming succeed. Well, they have none of those all of the advantages that you had all of a sudden become anchors that are holding you back from actually doing the right thing and that's why I

[11:28] You know, I always like to remind our company, Sierra, that, you know, the wave that we're riding of large language models and this next generation of AI is greater than any company riding it. And so, like, don't fight AI. It's going to happen with or without us. And if you go back to the Internet, if we were talking in 1995 or something, and we'd probably like search as a category, e-commerce as a category, digital payments, that stuff is going to happen. [11:58] been founded yet. I guess Amazon probably had around then. PayPal, probably not founded yet. The categories are obvious. The categories are like whether or not any of those founders existed, all of the would be winners. And it's the same now. It is the same now. So like everyone knows what's going to happen. And it's like you're competing for the privilege of winning. And so in a world where the technology is that [12:20] remarkably powerful, the strengths of the incumbents start to wither in the face of the technical change. And that's why you tend to get new great companies. The companies that are enduring tend to be created in platform shifts more than any other time. I'd actually be curious on this topic of there's these obvious things. And within AI, I would say, not to discredit your insight, but support, I would count as an obvious thing, like in a good way. It looks like it works. [12:50] you know, to a place, you know, at the right time, but other people did too. And so in some ways, I'm like, you have been playing both in a very blue ocean, you know, wide fields, like, you know, the incumbents are sort of like categorically different. And so like, it seems like inevitable that we're going to have agents doing support. And so there's that. And then on the other side, you know, a lot of other companies see the same thing. A lot of other people have been building. So before getting into the specifics, I'm just curious, like experientially day to day,

[13:20] does... [13:21] does your sort of operation of the company feel competitive or like wide open it feels competitive and it feels like a really big market um so it doesn't feel particularly demand constrained which is a really great feeling as a fellow entrepreneur it's like you don't get to so you feel like there's lots of demand and there's like a contest with each sort of situation yeah that's right the way it feels it feels like there's sort of too much capital available uh put another way there's [13:51] meaningful markets, it feels like there's sort of too many competitors that don't necessarily have strong differentiation. I think it's probably healthy, though. I think that, you know, there will be a culling, you know, just as the market progresses. But it does feel, you know, quite competitive. I'll just sort of give you maybe like a quick glimpse of the past couple of years. So we've had a remarkable growth in it. We closed 100 million in seven quarters, 150 million in eight quarters, which has exceeded my expectations. [14:19] But this past year has felt like an inflection point. So the first year of our company's history, we would often go in and be explaining to clients what an agent was. The term was novel and it was part of our marketing, explaining what an agent was. Number two, people would be talking about, hey, AI is maybe non-deterministic. They wouldn't necessarily use that word, but that would be what they would be describing. You know, how can we trust this technology directly engaging with our customers or consumers? What are the risks? [14:49] is [14:50] Clearly, we need this yesterday. I mentioned this to you earlier, but over a quarter of our companies have 10 billion or more in revenues. We're talking big companies. We serve most of the Fortune 20 as an example. And so these are big companies. They're coming in and saying we evaluate. We know what we want. We've heard of you. We've done all this evaluation. Here's an RFP. You know, like, let's go. And as a consequence, because the market has matured and, you know, by the illustration of the existence of things like RFPs, you end up in more competitive conversations.

[15:20] And then it's a question of like why Sierra? You know why Sierra and you know, I'm happy to talk more about that I mean obviously love to as well. I could say all the reasons are the greatest but you end up in this world But you're not explaining what the word agent anymore. You're saying here's why we're the right partner for you It's a very different conversation. Well, what I mean, so like, you know, they're like yeah, I'm bought in on an agent It's like why is it Sierra? Like what have you found is like the most important thing that you? [15:42] makes you win. So one thing we really did uniquely this year, the reason why over [15:46] quarter of our customers have over 10 billion revenue is we've tried to serve [15:51] more complex, more regulated industries. You know, we want, we serve most of the U.S. healthcare insurance market, as an example, and we serve U.S. banks, Spanish banks, U.K. banks, like, and these are companies that, you know, as you, if you know the industry, they're regulated by everybody. It's easy to make a demo on AI. It's why, like, you can go on X and just see a thousand demos and, like, demos are cheap, but making an agent sort of industrial grade is hard. And we've really uniquely been able to make agents that can actually have complex conversations. The other [16:21] that we do really uniquely is in addition to i think having a really easy to use product is we help companies move faster um uh we went uh live with cigna in in two months that's crazy which is remarkable yeah i mean how big is cigna it's a fortune 20 um healthcare company and i was on stage with suchin is who runs their ai practice there at the health conference and he was talking about this and part of that is like how can you show up at a like we're really great at ai cigna's

[16:51] How do you bring those two together to move extremely fast? And so for a lot of our clients, like the reason they bring us on is they can you help us move quick, move quickly. And that requires knowledge of AI, knowledge of business. And I think we sort of show up with a greater sense of maturity there. You mentioned like, you know, that the pricing scheme was one of the difficult things, you know, in the past. Yeah. You know, we don't have to like belabor it. But obviously, you know, going from, you know, just buying a license to a cloud subscription and now usage based is like the future. [17:21] What are you feeling as important as you have created and probably continue to iterate on pricing? What are the important levers for agent companies? [17:30] We do something specific at CIRA that I'm sort of an evangelist for, which is outcomes-based pricing. So it turns out in our industry, the outcome is usually well-defined. So in a service context, could the agent solve the problem? In a sales context, we do a lot of sales agents as well. Could it make the sale? You probably, your company has paid your salespeople commissions, right? Like that's where you can measure the outcome. You want to incentivize the outcome. The interesting thing about agents is they're autonomous or can be autonomous. [18:00] is measurable and trackable. What an interesting opportunity to actually [18:04] charge for that and if you look at the history of software like let's take advertising [18:09] We went from impression based ads to cost per click ads to now for mobile ads. You can do pay per install. At least that's my understanding. [18:17] And then you had enterprise software, you went from on-premises licenses to subscription-based software. And could outcome-based software be the next? And what's so neat about that is for a company...

[18:30] What an interesting and accountable business model. And I think there's some challenges to it because, you know, you obviously put some revenue at risk, but I don't think most advertising tech people would say CPC ads put revenue at risk. It's like the opposite, right? Because the closer you get to the outcome, the more valuable it is for the companies that are actually willing to invest in it. [18:51] And so my view is [18:53] To the degree agents have a measurable outcome, outcome-based pricing feels like the secular business model for agents. And I think it's quite both disruptive and I think a huge step forward. Why is it better than token-based? So like, you know, if those are like... [19:08] I guess sort of like the two reasonable options now. Why is an outcome better than token-based, even over the long term? Let's say you had an AI agent to generate leads for your sales team. What do you care about? [19:22] you care about the number and quality of the leads, right? And so you really don't care how many tokens the model uses. In fact, [19:30] it's not obvious to me that like there's a correlation between used tokens and leads generated uh and in fact in the same way there's no correlation in a sas product between their cost to serve and the quality of the product you know you could have a really good engineer write it or a really bad engineer write it though you really cause the quality of the product the reason why i don't think token based makes sense is it's charging for an input that is uncorrelated with

[20:00] And I think this is actually... [20:03] uh you know i'm a huge believer in applied ai but i actually define applied ai as can you describe your value proposition without without mentioning models because if you think about hey we can answer the phone and solve 80 of phone calls without human intervention with a csat score of 4.805 that's you don't mention models i mean models are an input to that but on output if you have to [20:33] It's not an application of AI. It's just sort of like a tool around AI. And I actually think that the closer you get to a business outcome, like it's actually you should charge for the business outcome, which is uncorrelated with tokens. And I also think it's almost a measure of are you actually an apply company? Yeah, if you can, if you don't have to talk about tokens, do you think that there will be markets either where things get so competitive that people have to price based [21:03] or maybe the other format for it would be if you can't describe the outcome cleanly. Like for example, coding, which we probably think is super important, obviously, it's like a little harder to say. [21:15] what the outcome is there versus like usage or something like that so like what are the conditions where like tokens do make sense yeah so i mean there's this old apple site where they had sort of like apple folklore kind of thing and i think there was this one boss at apple that made people thought of for him saying how many lines of code did you write and this engineer infamously wrote a negative number because he just like refactored a bunch of stuff it's my it's like the good analog historical analog for why tokens don't matter because it was he was it was his way of saying

[21:45] man like your lines of code has nothing to do with my value and he was doing it to sort of like you know piss off a middle manager to make that point but it's interesting is like in the world of software engineering people [21:57] Truly understand like the customers of those right now are software engineers who intimately understand these models. So there's a little bit of a the customer Product market fit. So it's a it's a nuance point, but I'll say like where I see it might happen. So right now [22:12] If you're evaluating a software engineering agent, a coding agent, you're probably comparing it to the cost of a software engineer. If you fast forward five years, you probably will be comparing it to the cost of other coding agents. So I think the second order effect as AI becomes prevalent is the reference point for its value will change. The thing I would say is that's true where you're thinking about a cost center. [22:42] necessarily apply. And if you go to my example of an AI agent generating leads for your sales team, depending on what you're selling, a lead is a lead of the lead. And you probably will value quantity and quality of leads. And there's a math equation. And that probably will remain independent of token costs, is my guess. And so I think a large part of AI is productivity and reducing costs. And there's a big part of it. But the other side of it is outcomes. And so could you imagine [23:12] five years where, you know, there's one coding agent that can actually produce something of greater value for your company. Will you value that? Or you just look at the token cause? I think probably you'll start looking for value is my guess. Will they all be the same? I don't know. You know, it's like, well, I was just reflecting on over the past year, there have been all these articles about has like AI progress slowed down. And then in our world of software engineering, it's been the opposite. Like every new model comes out, you're like, oh my gosh, it can write

[23:42] reasonably complex software. My theory of that is it depends on what you're testing. So if you're using ChatGPT for trip planning, probably haven't seen a material change over the past year and a half, because you reached sort of sufficient intelligence for trip planning a long time ago. If you're using AI to write Rust code, [24:01] Codex is like [24:02] mind-blowing right now. So I think one of the interesting things when I think about second, third order effects and the progress of AI is [24:11] You know, where you will you pass the horizon where like every model is sufficient in that task and then there'll be some things where like the frontier continues to move. Yeah, it's hard to imagine, but it's just like we're in a crazy time. Where are we at with support agents right now? Like, are there still edge cases, last mile things like. [24:28] that AI can't do still. [24:30] Yeah, we are, though. I imagine a lot of the [24:33] technical problems as opposed to product problems will become easier, but there's a lot of them still. So, you know, we at Sierra support most spoken languages in the world. And, you know, if you want to support Cantonese and Tagalog, most of the good voice models, you know, don't come from like the traditional Western model companies. Similarly, one of our clients is SafeLite Autoglass. So it's like roadside assistance. [25:03] car horns, background noise, kids talking background, you know, are actually all fairly hard problems to solve. And even in some of the advanced voice mode stuff, if you are in a noisy environment, it constantly thinks it's being interrupted and things like that. So you end up having to build proprietary voice activity detection, multiple speaker detection, all these other things. We develop all this technology because we need to be the best now. And I think we are the best now.

[25:27] And you're like, OK, that's probably going to be a commodity two years from now, one year from now. I mean, who knows? You have to do it because you need to be the best at every stage of your company's existence. And I think then the way we think about the world is we have a product which is called Agent Studio or Agent OS. And we're going to make the in three years, we'll judge us by our product. And right now we're probably our clients don't really put this way, judge by the technology. But if you go back to 1996, I remember when Netscape had a web server and Apache was new. [25:57] Like no one cares how you serve web pages now like it's a commodity. Yeah, yeah, but at the time that was what you sold and now you have increasingly higher order website building like Shopify. So I just think the agent market is going to take that progression. We're going from a tax centric sales cycle to a product centric sales cycle. It's interesting that you're. [26:16] obviously having to be the best at something that you know is going to get commoditized. Yeah. Which is probably not something, I don't know if you ever had to experience something like that. I mean, for that to be true, you just have to be in the middle of an insane rate of change. But that means you have teams who are putting a lot of their life force for two years into something that everybody knows is just for two years, but it still matters nonetheless. It's crazy. I mean, if you look at traditional, I'll just say enterprise software, consumers are a little different, but you think about you're building up this asset, your intellectual property is a fancy name for it. It's like, look at this, [26:46] platform that we're building. And we took so many years to build it, and it's got all these features. [26:51] And now you're like, I'm building this and I'm 100% certain we'll throw it away in the next 14 months. But I have to build it because if I don't, I can't serve the bank that has a big business in Hong Kong or whatever it might be where we need Cantonese support.

[27:06] That is the reality right now. I actually think [27:10] I've been thinking a lot about this, actually, just because I think it was Toby Lukey who sort of said something provocative around, you know, when generating the code is easy, it's almost like the system and the prompts that are actually the durable asset. You know, put another way, could you sort of terraform your software from scratch? You know, it's the prompts that led to it. I do think that is sort of the software of the future in a lot of ways, where how do you encode the infinite number of little product decisions that you made? [27:40] in code today. I mean, if you think about like a product requirements document versus the code, what percentage of the emergent product that comes out of is in code? Almost like 90%, like a lot of little detail. Yeah. Are in there. I think a little bit it's like software companies of the future and the products that they make are just going to take a really different shape in the future. And I I'm so excited to be a part of it. I mean, I think it's really fascinating. I think there's something, [28:06] really interesting about AI impacting the software engineering industry almost first and most. Because [28:14] We're disrupting the craft of making what we're building in real time, and it's fascinating. It's a fascinating time. I think there's a prevailing idea in tech. [28:23] that [28:24] AI is moving so fast that young founders have this massive advantage. And I mean this with no offense. You're not old, but you're also not the young. You're telling me I'm old. I got it. No, you're not the youngest founder. And you have one of the most successful AI startups there is. And it does seem like you've brought a lot of your previous experiences to what you're doing. But I can tell from talking to you that you also are just rethinking everything. And so I'm curious your own experience for yourself and for other founders you look around at.

[28:54] have the advantage? What does it take for more experienced founders to have the advantage? You know, I'm always... [28:59] Big believer. I don't know if it's a real quote, but some VC said, why was this founder able to conquer this market where so many others had failed? And they said, well, he was too naive to know it couldn't be done. And there's a certain element of that that I love because you end up with this kind of naivete that is actually sort of a form of principled first principles thinking [29:29] you know, dominates the market. You think there's a better, faster, cheaper way to do it. And because you don't have any of the hard-win lessons that can end up, you know, oversimplified analogies, [29:41] Keeping you from actually taking that leap, you can end up with, you know, Tony made DoorDash and didn't care about, you know, say, WebVans Monzo or whatever. I can't remember all the dotcom bubble companies. But I do think, especially in enterprise software, the experience that some of our team members bring, including the old man, me and Clay, bring to it really does matter. [30:11] We can go into a bank or a healthcare payer, a healthcare provider, a revenue cycle management firm, or a big telecommunications company and understand their business. [30:22] We're working with one large medical device companies consolidating 40 of their call centers into one. And we can have a discussion about the change management of doing that. And that's not really a tech problem, but it does require understanding business. And I think we always joke at Sierra, it's like the Venn diagram. There's a circle people understand, like next generation of AI, and people understand business. And we're like the company right in the middle of that, maybe the only one.

[30:52] there's that sort of infamous MIT study saying, all these AI projects fail. It's like none of ours do. And that's our value proposition. Like we can actually help you go live. And I think the experience has benefited us. - Yeah. I'm curious like if you can point to [31:06] what has created the lead you have so far. And obviously, I know you're just getting started, but at the moment you do, you know, you've pulled away in a big way. And I'm sure there's a lot of just like daily blocking and tackling, but I'm curious if there are any like foundational decisions that you've made or strategic approaches that, you know, over the last couple of years, you look back at and you're like, that was pretty essential to make this happen. I think there's two almost independent [31:31] um areas of investment not not they're not independent but they're like uh very different one is the product and one is our sort of our go-to-market and partnership model and they're both really intentionally built on the product side we've tried to balance ease of use and extensibility because when you serve really large companies with very comp that have been around for 200 years you know you need to work with mainframes you need to work with a thousand [32:02] And so that's why you tend to have, you know, most, I'll say, enterprise software that's designed for larger companies tends to be quite extensible. Often that extensibility comes at a cost, which is, is it easy to get up and running? And so as a product designer, like one of the things I've just spent a lot of time thinking about is like, we're trying to have our cake and eat it too. Like, can you go live in two months and still be maximally extensible? And I'm really proud of the product that we've built. And some of that is born from experience of what does extensibility mean?

[32:31] of what it means and have been able to accommodate [32:35] like some fairly exotic deployment requests and still do it fast that's really unique the second thing is our go-to-market and partnership model because we knew when we started the company we wanted to work with the largest companies in the world not only but we want to be able to work with largest companies in the world and i focused on that and as a consequence we just have a really unique partnership model um there's sort of a fashionable thing to talk about forward deployed engineering in silicon valley we don't call it that and it's a very unique model [33:05] technology like most of our clients build and maintain their agents themselves. It's pretty easy to do, but we show up and we help you be successful. And so it's like, we'll just show up like we're not going to let you fail. Like and I think that is a very different because we have this outcomes model outcomes based pricing model. We don't get paid unless it works. And so how much of that is technical versus like, you know, [33:26] change management? It's a mix of both. I don't know if it's 50-50. Do you know it as two people or it's one person who does both? We have a mix of roles. We've sort of evolved that. We try to hire really technical people in all roles, though, because part of our secret is we want [33:42] we want to be like be your trusted partner in AI. So you want the person who is working with you every day to be the most knowledgeable AI person, you know, like a forward deployed change management engineer. Yeah, exactly. It's crazy what we're doing. And so and what's really neat about is if you're like a really talented technical person who wants to go transform an industry, you can do it at Sierra. I mean, you can go in and like we're working with most of the health care insurance companies like you want to change health care costs. And, you know, like what a cool vantage

[34:12] You said that it's not just support agents now. Yeah. It's like, what else are you finding shoots in? I'll give you one of my favorite relationships with Rocket. So based on Detroit, remarkable story. Their founder has done more for Detroit than I think any one person's done for any city. Just like remarkable company. But they own Redfin, which is a home search site, Rocket Mortgage, which is like the number one consumer mortgage originator in the country. [34:42] recently as well and you can go to redfin.com and use an AI agent to search for a house you can go to rocket.com and finance that house with an AI agent and then you can with the acquisition they do this mogu service firm you can then when you're servicing your mortgage you'll talk on the phone with an AI agent as well so like everything from finding a house yeah to originating the mortgage to servicing that mortgage I think it's pretty cool and like they have an amazing CTO name Sean Mahotra like pretty visionary and I love their CEO [35:12] ruin too, but it's like everything from finding a house all the way through servicing. It's kind of what we believe a lot of businesses will do is like look at their entire customer life cycle from, you know, I'll say purchase consideration, which is a fancy way of saying like browsing. I think homes are probably one of the more considered purchases that you could do through executing the purchase. They're having issues with it all the way through, you know, retention. And for a lot of, for example, a lot of our telecommunications customers, their AI agent

[35:42] So like you've probably negotiated your cable bill at some point. Yeah, probably. And so our agents are doing billions of dollars of negotiations for everything from, you know, satellite radio subscriptions to cable television subscriptions. It's pretty cool. I mean, it's like really, you know, over a billion dollars of mortgage. Basically, just like all transactional communications. Eventually, the way I think about it is. [36:07] website is a technology, but your dot com, the one with your brand at the top is your website. We're sort of doing that for agents. It's sort of like agents will do a lot of things. The one with your brand at the top that your customers go to, whether it's buying or servicing, we would like to help you make that. And I think it's an interesting, as agents go, it's often interacting with other agents, right? If you think about a home and auto insurance company, you may have [36:37] You know that's quite complicated. So our agent that's having the phone conversation when you're on the fender bender will interact with that But it is almost the intersection of all of the that technology because it's sort of your your front door and our whole hypothesis is Every company needed a website in nineteen ninety seven every company needs an agent in 2027 and like we want to be that that company What's the nuance about like agent builders though? [37:07] yeah I mean [37:08] I've been surprised how many large-income enterprise software companies, their first foray into AI was

[37:16] you can an agent building tool it just feels inevitably to be a commodity in my mind because you may be making a website was hard in 1995 but today there's like a million ways to make a website most of them are open source so you have like cool companies like purcell which i love but it's not like there's a huge market for this stuff um and and in practice i think the same will happen with agent building um i think open ai will have a great tool probably all the foundation model companies [37:46] is like Langchain and Langgraph. Mm-hmm. [37:50] The idea that you you have the right to win there, I don't know if anyone has the right to win there just because it's just a technology. It's a horizontal technology. And I just believe in open source and it's just going to become a commodity. So my belief where there's value is really going to be an agents that do things and you'll hire those agents and purchase those agents for what they do. So I believe in companies like Sierra. I believe in companies like Harvey. [38:20] have an agent that will do an antitrust review. You know, I think there'll be a finance agent that audits your financials. There'll be one that helps you onboard a, you know, supply chain vendor. There'll be one that, you know, if you just think about onboarding a new vendor, it's like there's a procurement process, the legal process, there's a contract review process. Whether or not it's completely autonomous or human in the loop, all of that could be augmented with an AI. And I'm like, that's a product. Agent building is not a product. It embraces a technology.

[38:50] aside from being the founder of Sierra Ural. So, [38:52] on the board of OpenAI. You're the chairman there. I wanted to ask you specifically about Codex. Like over the last, you know, couple of weeks, it's been unbelievable. It's like, you know, a curtain just came down. Did you expect this? Like, did you think that what has happened here was going to happen? Or like, when did you start to have an inkling that like code was going to go vertical like this? I'll say yes, I expected it just because, you know, being on the board of OpenAI, we talk a lot about it and all the labs, Anthropic and OpenAI in particular, [39:22] coding agents to help build AI. And certainly, building an AI researcher is an important part of building an AGI lab. The weird part about, for me, as someone who is a software engineer-- [39:35] I didn't feel it until I used it. So you like you can talk about it all the time. And then like the first time you one shot something, it turns out like really good and not like slop, but like really good. It's an emotional experience, I think. I mean, for me, it was it was just sort of like, holy shit. [39:55] Like, this is real. Yeah. As you said, it's really over the past years. [39:59] three months that has felt really materially different to me and i've been thinking about it a lot um i was thinking about the past 20 years of software engineering um i remember the first time i worked on the engineering time that had real ci cd where you'd check in code and it would just automatically end up into production and i remember how i'll just like if you've ever worked

[40:29] it's completely different because to have something that can safely go from commit to production, there's so many things that have to happen to make that work. You end up relying a lot on testing. So both unit testing, integration testing and canary testing, because the last thing you want is someone clicking a button and taking down the service. And it's almost impossible for a team that is doing manual releases to convert into CI, like true continuous delivery, because there's so [40:59] with that. It's easy to start that way and very hard to work. So I've been asking myself, clearly in three years, we're going to, like if we were talking, we could talk about what are the best practices to set up a software team that's optimized for this technology. We'll know what those [41:17] best practices are. And right now we're just figuring them out in real time. And like, my hypothesis is the companies that figure it out first will move the fastest. Yeah. And, and the other part of that's the companies that don't will move much more slowly. It's fascinating to me. Uh, and Andre Carpathia, he had a really interesting post about this too. Like, I think a lot of folks are sort of like in deep here, have been thinking about it and it's fun to see the industry you love sort of flipped on its head and yeah. Well, it's interesting. Cause like, [41:47] like you know software engineers on one end and then like say somebody who's like you know [41:52] in some part of the country where AI has not yet gotten fully extended. There's a wide gap in people's current...

[42:00] sort of comprehension of like what AI is going to do. And so I think, you know, it's like it's a little bit unknown. Like, you know, there's a lot of blog posts going on right now that are breathlessly saying like it's all over. I think, you know, I'm probably more in the camp of like maybe software is like, I don't know, people, you know, use the word software is solved. I don't know if it's that. But I'm curious if you have a view on like if Codex and Clubcode and sort of like the latest in coding, is that going to change the way companies are built? You know, like one [42:30] that there's going to be my brother, you know, these 10 10 person billion dollar companies. You know, is that are we at the precipice of that? Does that make sense? Are there other changes? Like what's going to happen now? There probably will be a 10 person billion dollar company, but I don't necessarily think will be the norm. And the reason for that is competition. If you imagine like the mobile phone market in the United States, there's three main competitors, Verizon, AT&T, T-Mobile. [42:55] And they're all competing for a fixed pie of mobile subscribers. And it's why it's extremely competitive. There's promotions, there's ads. They can't make more of us. They can't make more of us. They can build up their network. They can do other pricing and packaging. And it's a really complex business to run. [43:13] All of them have access to AI. [43:15] every single one so the idea that you could deploy ai and you know not have to do things you're doing currently because of ai is probably true but if any one of them figures out a way to use a person to gain market share against the other one they're going to do it yeah and then as a response their competitors will do it too and that's how you know we spoke about this earlier but it's the reason why when automated teller machines were introduced to banks the teller job went away but

[43:45] and no fewer people in those bank branches. And it's because, I don't know if it was JPMC, or someone figured out, hey, if we put financial advisors in there and other things, we can actually make more revenue per branch. My personal take is in a competitive market, and that's the key, by the way, you need competition. So people can't just pass the cost savings on the shareholders or dividends. The second order effect of the efficiencies of AI will be investment to compete, lower prices, or customer acquisition, or whatever it might be. - So we want a fewer engineers per company. [44:15] they'll be way more productive, and so you just end up with way better software. Or you might have fewer engineers and more of something else, or you might have more engineers. You know, I don't, I'm not sure, but it's the idea that like it will be what it is today, but just more efficient, I think, is like a lack of imagination, in my opinion. The interesting thing, though, is the other part of this [44:34] Software engineering does feel special and I think people extrapolating too much from software engineering or it's a bit simplistic. You like the same thing might not happen to every other function. I'll just be really simple about it, which is finance and software engineering might be limited by intelligence, meaning they're largely digital. They are largely like manipulating sort of digital things to and you could imagine AI automating that most of the economy isn't. [45:04] digital, exclusively. So if you need to ship something, a T-shirt from Vietnam to here, yeah, you could automate some of that stuff. But at the end of the day, that cargo ship still needs to be in the water.

[45:17] And I always bring this up, you know, like just imagine you run a pharmaceutical company, you know, you can think about, you know, how to make a therapy. You probably need a wet lab. So, OK, well, that's intersects the real world. Maybe you could do robotics, but then you need a clinical trial and then, you know, so just a lot of the economy is like real. And so it definitely will change the way companies are built. But I think when people say everything will be 10 people, it's like maybe just the stuff that lives in bits. [45:43] Yeah, that's right. Which is a lot of the economy, but not the economy. Yeah, I mean, you know, it's easy to like to talk about this, but you're right. Like if you just like move around the physical world and you get off of, you know, this podcast and, you know, this computer I'm sitting in front of all this stuff and you got into the world and there's like, you know, trucks moving dirt around and people who need a building that has lights in it and all. There's like a lot of physical things. [46:13] I think you're probably right. [46:17] And so, you know, like robotics will have a big impact as well. But I think [46:21] People are thinking about this a bit simplistically is my my take and I think Intelligence is clearly on the cusp of going up exponentially, but it doesn't mean adoption of like that can't be absorbed by the economy and [46:34] perfectly exponentially. And so I just think people are a little bit simplistic. Do you think there's any cognitive things that are [46:41] immune from intelligence like Dylan Field when he was on this podcast gave an example of like Brat Summer as something where he was just like that would have been such an insanely hard call for an AI to make. And you need so much context and taste and opinion. You know, where my head was going is, OK, so coding is, you know, whatever's happening there is happening there. But what about like brand or storytelling? Like I'm kind of asking you this both as an operator and as you know,

[47:08] somebody who's very deep with open AI, do you think that these other parts of intelligence also... [47:15] you know, go the way of AI? I don't know if... [47:18] taste is necessarily related to intelligence. You know, it might be, but I've got three kids, including a 16-year-old and a 15-year-old. And when they decide what they're going to wear to school, I don't think they will consider chat GPT's opinion. They care more about what the person in class next to them is wearing. Yeah. [47:40] Similarly, if you go to the most elite, competitive, college preparatory... [47:48] school or the worst school in the world, there's always going to be the smart kid in class and the dumb kid in class and the strong kid and the fast kid and all these other things. And like it's all relative and it's all very local and it's all very human. And so I think the idea that because AI is smart, it takes something away from us as humans. I don't necessarily subscribe to. I don't, you know, you all see these things that go around online where people are sort of lamenting older technology. [48:18] like the bicycle. - Yeah. - And we've been weaker than machines for my entire life. - Yeah. - And I don't, [48:25] I don't think it like it doesn't make me feel like weak as a person. And I think we this for the first time we have computers that are going to be more intelligent than us. I think there will, you know, the emotions I had about codecs writing code that was high quality. It wasn't experience because, you know, I might have some of my identity tied up in that task. Yeah. And the next day I woke up and I'm using it as a tool and I can make better software. I'm like, this is great. Probably actually like a good like self-actualization anyway to go through that.

[48:55] I think this is I think people's vocations and their identities are often very intertwined. But I think once you absorb the technology, I don't think it's actually your identity. Yeah. And so I think I actually am quite optimistic that we will be human. We all be status seeking animals. We all compete for the real estate here in San Francisco. And even though our standard of living will go way up. Yeah. We will all be jealous of people still. We will all compete. And as a consequence, I think humanity will be just fine. [49:25] That's my view on it. And I think it's just hard to imagine. But it doesn't mean it's going to be catastrophically bad. I just think it's actually, I think it will be largely good for humanity. I have a friend who believes that as this kind of progresses, you know, we're already, everybody's already completely addicted to their phones and it's a disaster and whatever. Now you have all this AI happening. A friend of mine was saying that he basically thinks that it'll actually become a status signal to become increasingly offline. And I'm like, actually, that might be an interesting call. [49:55] and like [49:56] intelligence will get so good and then people will sort of just be like enough of all of this and like hopefully there's a big screen time reduction you know and it's like so like parents were revolting on social media like about social media for their kids and like a bunch of schools and all the parents like nobody take a phone like everybody agree to it so i think that'll be an interesting thing of like does humanity like does it is there like an essential humanity that like [50:17] gets sharpened i hope so actually one of the things you know i i love the iphone is one of the greatest inventions of uh this century i hope we're not staring at a glowing rectangle it can't be the right way to do it and and you know now that ai can talk to you and human computer interfaces like so this is my point i actually think hopefully humanity can become more self actualized yeah you know as a consequence of this and that is the

[50:42] a purpose of technology. So, you know, just like the Industrial Revolution had Luddites and globalization led to job loss in the Rust Belt of the United States, but certain goods got less expensive and other parts like these, there's not going to be no issues. I think it would be callous and insincere to imply otherwise. But I think it will largely just really accelerate humanity in a really positive way. And I think that [51:12] for me and I think for like if you're thinking about how does this impact me is like have a more flexible view of your own identity like the what how you do it every day doesn't define you I was like the metaphor because it was so obvious before and after imagining being an accountant before Microsoft Excel and after Microsoft Excel yeah so much of the act of an account was like adding up numbers and things you know and now it's like building a model and it's not like what you did like the value you provided didn't change but actually the act of doing it is completely [51:42] different. So I think it was just like a lot of us are just going to go through that in a very compressed period of time. And it's OK. It's just a little anxiety. Yeah, it makes sense. My last question about AI, there was a shot from Anthropic at OpenAI around the Super Bowl commercial about the ads, which is there were good ads are funny. But then I think sparked like a debate around sort of like the whole topic of like, what is the role of these foundation labs?

[52:12] business model? What are the tradeoffs of all of this? You've obviously like, you know, you have experience with social networks and a lot of different pricing, you know, models, you know, open AI, well, you know, you know how to consume AI. So I'm just curious how you think about this and like, what is the right thing when you consider like a lot of these dimensions? I'm very optimistic about ads. [52:33] done in sort of a tasteful way. You know, I started my career at Google, I think I arrived like the day AdWords came out. So it was just interesting because when I started there, you'll laugh at this, but like everyone in my family, when they found I was working there was like, how do they even make money? And laugh just because I was I think I listened to the acquired podcast is literally the most profitable business ever created. But as a consequence, you know, Google is widely available for free for people who want to use it and has created an economy around it for [53:03] or demand fulfillment advertising. I think there's reasonable criticisms of advertising, you know, if it starts to get in the way of the sanctity of what the AI is recommending you, which was sort of the backhanded implication, but I just think it's not true. And so I actually think if ads are clearly labeled and not getting the experience, [53:25] I think it's really aligned with the OpenAI mission, because our mission is to ensure artificial general intelligence benefits humanity. Obviously, the most important part of that mission is safety. But after you get back the Hippocratic Oath, first do no harm, the job of a doctor to cure you. So then after you say, OK, it's safe, how do we widely distribute it? And I think we have an obligation, being a mission driven, you know, I'm the chair of the foundation and on the PBC board. Yeah, like. Yeah.

[53:51] Our mission matters and being able to offer it for free widely is a huge part of that and we need to be able to to afford that Yeah, I think it's not only I just I find it [54:02] inauthentic. Like I'm like, this is an incredible opportunity to provide this at scale to society. And I think the idea that it will somehow take the experience. It's funny, you know, like I grew up in like a suburb of St. Louis and, you know, so it's like a whole different world than like, you know, what we're in now. And it's like when I think about like, you know, people, you know, that I grew up with or, you know, from just other parts of the country, 20 bucks a month is a lot. And I think, [54:32] a month on stuff but they really want these services like you know if the whole world had to pay for google like that'd be a worse world like it's really good that everybody has access i just think it's important we do it well yeah yeah people want good ads like i like good ads like i would actually if people bring me the right product i'm like that's really nice this is the other part of it is like you want businesses to be able to grow from scratch there's such a purpose of it it just needs to be done in the right way so i i find the discussion not not particularly

[55:02] chosen to sort of like finance the company. And I guess I'm curious about three parts, which are how you got started and, you know, working with Peter Fenton and then like what you've done since then to date and what's been important for you. And then I'm curious, just like, as you think about the future, like what's important to you as you think about other partners or capitalizing. And, you know, I'm asking just because this is a podcast has a lot of VC in it. So I got to have a little flourish. Yeah, totally. We have three members of our board, which are sort of represent [55:32] Peter Fenton from Benchmark, Ravi Gupta, who just left Sequoia, though he's still a venture partner there, and Neil Mehta from Green Oaks. Just a fantastic group of people and chose them all both for the firm and the person. But notably, like Peter, I've worked with both my previous companies. So, you know, our first round of financing, I didn't talk to anyone else and introduced him to Clay, my co-friend who hadn't spent time with him. And we talked once. He's [56:02] sent me a term sheet. I signed it, no edits. And it was like a very much a trust relationship. And it is interesting, like one of the things I really have appreciated about. So there's some downsides to Silicon Valley and how insular the community is. One of the great parts, though, is just like the relationships you can forge over years. And for me, it meant Peter and I could sort of start on third base just because we've worked together a lot before. And so you just don't end

[56:32] the funny business in the boardroom is just like, let's get to work. And it's fun. It was fun to, you know, sort of get the band back together there. But the fun part for me is I had never worked with Ravi nor Neil before. And like Clay and I just it's like it's just it's just a great board. Yeah. Like it's like people we seek out advice from as opposed to. [56:51] people we report to every quarter. It's amazing. How do you think about because you're both like, no, like, you know, when when opening, we won't go back to the story, but like, you know, when opening, I had it's like, oh, my God moment. Like Sam was like, you know, right. You got to like you're like the board member and then you've also got a board that you're so you're you're in both roles at once. How do you like make the most out of the board? [57:12] you know, obviously you've got these particular relationships, but like what do you expect that relationship to look like? First, I really like written documents for boards over presentations, both as a board member and as like a founder of a company, because you end up letting people synthesize information ahead of the board meeting. So you end up with more substantive discussions in the boardroom. I've done this for the last two companies I've started, and it's just been amazing. [57:39] Great to send out a board document. Sometimes people will comment ahead of the meeting, but I actually think the main thing is it's been read and it's been read ahead of time. And then you end up with a meeting about the actual meat and potatoes of the topics. You're not like staring at a bunch of sales numbers for the first time.

[58:09] - The process of the writing? - The process of the writing is a process of clarifying your thoughts. And so for Clay and me, this is a process by which we synthesize what's been happening. And you know it, you talk about it, but to actually write it and write it eloquently and concisely is incredibly important because it's essentially a way of, you know, it's like, what's that famous line? If I had more time, I would've written a shorter letter, like spend the time. 'Cause that's actually how you can show respect to your stakeholders that you're thinking about the strategic issues going on in your business. [58:38] And the last thing I'd say is board members aren't sort of single issue voters, but everyone has their strengths. And, you know, at OpenAI, we've recruited a pretty diverse set of skills. Zico Coulter is a professor at CMU who specializes in, among other things, jailbreaking. So just like one of the experts on some of the more subtle safety aspects, Nicole Seligman was a great attorney. And, you know, she's an expert in a lot of like legal issues. [59:08] investors too is find people that your management team will want to go to for advice obviously the audit committee chair and your cfo have a really unique relationship but you really want folks like who's your head of sales going to go talk to do you have someone who's like kind of been there done that because you want them to have that kind of like i always think of it as like who are the advisors you want to surround your management team well and i think a functional board really has those relationships and then when you're in a board discussion you have all these board members who [59:38] company, but in a really valuable kind of targeted way. So I like to think of the board as a collection of people. Don't look at the individuals. The whole should be greater than the sum of his parts. Anything this year you're particularly excited about that you can share? I think the real exciting part is going to be

[59:55] adoption and regulated industries. I think we we're moving beyond like the early adopters to everyone. And so I think we have we talk a year from now. You're doing the hard stuff. It's gonna be like the really hard stuff. And [1:00:09] If you want a hot take, I think [1:00:11] My intuition is regulators will start asking for agents. The idea that you have a human set of controls over a regulated process will start to feel like a risk rather than the risk being AI. And that's my I don't know what happened this year, but I think that. Well, I'll call you in a year and we'll do take two of this. That's great. All right. Thanks so much for this, Brett. This is great. Thanks for having me.