Back to Index

State of Startups and AI 2025 - Sarah Guo, Conviction


Transcript

. So first question for you. What is definitely happening by the end of 2026? AI agents ship code directly to prod in your environment, right, not in some playground. Voice AI replaces text for most business communication. Inference cost drops below a cent per million tokens. Or Wall-E, like we're all chilling.

Any of these? First one, ship code directly to prod, okay, this is a hopeful set of engineers. All of you want to get rid of your own jobs, I love that. The good thing is I also don't have Internet, so I can't look at my next question. No, it's going to be good, it's going to be good.

Do you want to present from your phone? Oh, no, I was going to go through poll questions while we're trying to do AV setup. This one though? It's going to be good. Yeah. While this is happening, I'm actually just going to introduce myself so we're not wasting the time.

So, my name is Sarah Gua. I helped start an AI-native venture fund. It's called Conviction. We got going about two and a half, almost three years ago now, just before the starting gun of Chad GPT. As always, in technology, investing most of life, it's better to be lucky than right.

Hopefully, you can be a little of both. And the point of having a new venture firm, I worked at Greylock. It's kind of a traditionalist venture firm, a great one. My partner, Mike Vernal, used to work at Sequoia. You guys have probably heard of them. Was that we think, like, actually, you know, at risk of sounding like those people, this time it's different, right?

That this is the largest technology revolution that we get to be a part of. And that there's so much change in the technology, the types of businesses you can build, the product decisions you make, what challenges these startups and big companies face that, you know, maybe there's opportunity for, like, a startup VC as well.

And so, you know, I'm thrilled to be working with, like, really interesting people in the industry so far. Mike and I are investors in companies like Cursor, Cognition, Mistral, Thinking Machines, Harvey, Open Evidence. So a mix of base 10, like a mix of infrastructure, model, and application level companies.

And, you know, one more, are my kids coming up yet? Okay, cool. One more just observation from the last two and a half, three years of doing venture. I was an investor for about 10 years before that, is I have never seen the, like, just the uptake from users that has been possible in the last couple years.

I'm sure all of you have experienced that. It is not trivial. You know, AI product and AI engineering, and this is kind of the theme of my talk, so I'm sorry to give away the punchline, but it's quite a bit harder than people had hoped. But the value creation is massive.

We see companies going from 0 to 10, 50, 100 million in run rate very, very quickly, faster than we've ever seen in any technology revolution before. And I get asked a lot, like, where are we in the AI hype cycle? Is the winter coming? Is this, like, infinite AI summer?

And I would say, having actually been an investor or an operator through a macro cycle at this point, like, I try to pay very little attention to what the marketing world is saying, or even what the markets are saying, right? Because, you know, if you're an operator or an investor, maybe you care about what the stock price does every day, but really you want to figure out if the company you're working for or starting is going to work long-term, right?

And if the products are going to work long-term. And the things that I get most excited about are seeing, like, crazy usage numbers. Okay! Whoo! Thank you, amazing AV team. Okay, I'm going to go real quick. Where are my presenter notes? Okay, we're just going to keep going. It's cool.

It's cool. So, I want to talk really quickly about just a few things today. I think we lost a little bit of time, but let's say let's talk about capabilities, what we're seeing work in the market, and then maybe some advice on, like, what to build, if those are, you know, a question you're considering.

I think the shorthand that we're going to use in this presentation is, like, cursor for X, right? And I do think that's a really massive opportunity. The first thing in capability for this past year is clearly reasoning. Reasoning's a new vector for scaling intelligence with more compute. The labs are really excited about this because they get to spend more money and get more output.

But we should also be really excited about this in terms of unlocking new capabilities, right? If you just put aside how it works, it's a confidence-boosting implementation detail. But we should expect more capability. You're unlocking a new set of use cases, like transparent, high-stakes decisions where showing the work matters.

Sequential problems, problems where you need to do systematic search. I think this looks like a lot of problems that we're excited about and face in knowledge work every day. As you have just seen demos of and I'm sure are working on, given reasoning, people are really excited about agents.

To put a -- you know, I want to do, like, the Steve Ballmer impression that's, like, "Agents! Agents! Agents! Agents! Agents! Agents!" But I -- you have to give me more than 12 minutes to, like, get that sweaty. But, like, the non-marketing definition that I think of is it's software that -- it takes some set of steps.

It, like, plans. It includes AI. It takes ownership of a task. And it can hold a goal in memory, you know, try different hypotheses, backtrack. It ranges from super sophisticated to super simple. Some of the tools that you might use to accomplish a task include other models or search.

And largely, it's just, like, AI systems that do something. And that's not a chatbot that looks more like a colleague. And, you know, one thing that I think we have a really unique vantage point on is we back a small number of companies at Conviction, but we also run a grant program for AI startups.

It's called Embed. We get thousands of applications every year. And it includes, like, user data and revenue data and, like, really amazing people. And the number of agent startups has gone up 50 percent over the last year. And a lot of them are working. Like, we do see stuff that's working in the real world.

And that's super exciting. Other modalities are progressing, too. I'm sure a lot of people are using voice, video, image generation, even beyond, you know, Studio Ghibli. But you have companies like HeyGen and Eleven and Midjourney that are rocketing past 50 million of ARR. These are real businesses now. I want to see if I can quickly play for you.

They told me to express myself. So I did. They told me to express myself. So I did. Now I'm banned from three coffee shops. Hands can hurt or heal. That's the difference between chaos and creation. So if you're wondering where Q3 is headed. So if you're wondering where Q3 is headed, here's the thing.

Consistency always beats urgency. We've got the projections ready. And let's just say, it's looking solid. I would definitely recommend it to anyone. I would definitely recommend it to any. So I think, like, if you just are looking for artifacts of improvement, this is from a company called HeyGen. You can make clones of yourself, of fake people.

And it's like you have gestures and expressions that reflect emotion and content now, right? So these models work together. And like, I don't know about you guys, but looking at that last gal, like, I feel influenced. I don't know what the bunny is, but I would buy it. And so I think, like, huge swaths of the economy are going to be affected by this sort of multimodality.

Some investors or operators would say multimodality would just be for niche verticals that enterprises don't have. You know, your average enterprise doesn't have that much voice, video, image data today. But I think that changes, right? When you can do stuff with this data when it is structured and understood, there's more reason to capture it.

And I think of, like, how much video do all of us watch every day? It's one of the highest bandwidth communication methods, and we're just going to use more of it. We think voice is where we're going to see applications first in business workflows, because it's already a very natural communication mode.

So everything from medical consults to lead generation, places you already had business voice, you just couldn't scale it before. I think that's where we're going to see it first. But as these other modalities become more controllable and also less costly, we should see all of them. I think it's safe to say you can expect capability improvement in every part of the model layer, which is really exciting.

A lot of people are talking about the data wall or, like, the end of AI summer. But for anybody who's building applications, I'm at least to tell you one person's opinion is not coming. And then, usefully for all of us, that market for model capabilities is getting more competitive, not less.

Sam Altman himself, I think, said it best. Last year's model is a commodity, which is a scary thing for a model provider to say. Because last year's model is now pretty damn good, right? The numbers tell the story. GPT-4 went from $30 per million tokens to $2 in about 18 months.

The distilled versions of that are, like, now 10 cents. So we can really use them very broadly. If you look at this chart, green is Google, yellow is Anthropics. You see, you know, it's a real mix. This is data from Open Router. So thank you, Open Router for that.

But you really saw Claude cut into OpenAI's market share and Google come roaring back with Gemini. This data is obviously a little biased because a lot of people just go direct to OpenAI. But if you're into multi-model, there really is a mix. And you do have credible new players like SSI and Thinking Machines, some of the best researchers in the business, with orthogonal technical approaches entering the fray as well.

And I'm sure many of you have experimented with DeepSeq, coming out with releases of both base and reasoning models that are reasonably competitive with a claimed fraction of the training cost. Like, we should just assume that open source will do as open source does. And we can rely on the model market to compete for our business, which is really exciting.

And so the view is plan for a world that is multi-model. Tools like Open Router or inference platforms like Base 10 help that, and I think, like, be comfortable with that. I am. Okay, so we have all this capability. Let's shift quickly to the application layer. We have to start with Cursor.

A million to 100 million of ARR in 12 months and half a million developers. I assume all of you. Zero sales people to start. That's not growth. That is a killer application. Cognition, which started with more autonomy, is already the top committer in many companies. Feeling a little threatened, but also excited because recruiting is hard.

And then Windsurf, who's on a tear itself and really beloved, is being acquired by OpenAI for three billion dollars. So we know for sure that the labs don't think that they can just, you know, steamroll everyone. Right? Lovable and Bolt hit 30 million of ARR each in a handful of weeks, helping non-engineers vibe as well.

So, you know, our ranks are expanding. And I think it's useful to just, like, analyze a little bit why code was first. Fundamentally, it is text with its, like, logical language with structure. Right? So much of coding is sophisticated boilerplate. Like, we all love engineering, but some of it is, like, craft work, not new algorithm work.

You don't need AGI to write, like, an API endpoint or a React component. Second, you have deterministic validation. You can automatically check if code works. Run tests, compile, execute, do things developers would do. And third, researchers believe code is crucial for AGI. Right? So they poured resources into it.

And code became a key benchmark and a training priority and an area for data collection. But I think the last point is the money point to me. Engineers built tools for engineers. They understood the workflow intimately, and that made all the difference. And that last part is the playbook for every other industry.

I'm sure people are building things that serve beyond engineers. And I don't think the winners will just be AI experts learning those domains. They'll be customer-centric, like, problem-centric builders who understand AI and then redesign workflows from first principles around manipulating those models. And so I think that's really the opportunity to build Cursor for X.

Let's think a little bit about what that means. Cursor is not a single model. You know, one model's doing diffs, one's doing merge, one's embedding the files. They manipulate and package up the context. They prompt the models very skillfully. They let engineers avoid repetitive tasks and standardize with things like Cursor rules.

And then if you're using Cursor in a team or even yourself, regularly retrieval accuracy gets better the more you use it with coverage and freshness. And so all of this happens in a UX that makes sense. Right? Like, I use VS Code. I'm familiar with it. My shortcuts work.

And they make it safe to say yes. Right? Like, green for add and red for subtract makes sense. I can scroll through it. And it's fast enough that I don't get frustrated. So my view is Cursor, if it's a wrapper, it's like a very nice, thick, perhaps 14 or 15 billion dollar wrapper.

Right? It's like if your burrito was 80% wrap and 20% fill, but you got to choose the fill and there's like an empty, like an open market for fill. Right? And so where's the value now? It may not be in the protein. It's kind of in the company. So, like, if we try to generalize that recipe a little bit, if you are building a generic text box, like, unless you're just like learning to do this, please don't.

Like, OpenAI already won that. Or it's just not very valuable to do so. Your domain knowledge, your workflow knowledge can be the bootstrap. If you already know what users in your industry need, don't make them explain it. Build products that show up informed. They collect and package context automatically, including from other sources, not just natural language.

Present it to the models. Use the right models at the right time, now known as orchestration, and present the outputs to the users thoughtfully. Right? So I do not think this is the end of the GUI. I think you can capture and enable workflow with these models. And all this requires taste and a ton of work.

I'd argue that, like, some version of this recipe is much of the work each of us is going to do. So don't listen to the labs from a user experience perspective. The prompt is a bug, not a feature. I think it's like a stepping stone. Don't make me think as a user.

The best AI products, they feel like mind reading because they are. There's enormous headroom in building these products, and I think that's really exciting because that's what most of us in this room have alpha on. What is a software company if not a very thick workflow wrapper most of the time?

That was true in 2015. It's true in 2025. Besides code, where might you go apply this? We think the opportunities to build value around the LLMs exist in every vertical and profession. But here's something counterintuitive. Beyond coding, one of the things that I've been surprised by is that the most conservative, low-tech industries seem to be adopting AI fastest.

We call this the AI leapfrog effect internally. These are three portfolio companies. They're working. Sierra resolves 70% of customer service queries for their customers. They serve people that you guys use, like SiriusXM or ADT. Harvey is two years in, well over 70 million of ARR. Its AI is essential now to being competitive in the legal industry.

There's a company called Open Evidence, which helps doctors stay up to date with medical research. You have to be a clinician to use it, but you give it your medical ID number, and you can do intelligent search against medical research at the point of clinical decision making. Today, it reaches a third of doctors in the U.S.

weekly, and the average user uses it daily, right? And so I think there's just examples of huge value beyond ChatGPT. These are companies that know their customer and solving real problems. As a piece of trivia that you may or may not know, Brett at Sierra is the chairman of the board at OpenAI.

OpenAI was Harvey's seed investor, and if these people are not fretting about thin wrappers, I suggest you don't either. Okay. Finally, I'll make an observation. A lot of people are excited about full automation. Now I'm sweaty enough, so agents, agents, agents, agents, agents, agents. But when we analyzed the applications to embed, I said, you know, it's gone up to 50 percent, you know, doubling applications for agentic startups in the last year.

I think some people think copilots are yesterday's news. They want to get to the endgame, right? Like, you know, your colleague and AGI. But in terms of what works, like the data on what's driving revenue, I think copilots are still really underrated. We see a whole spectrum of how much automation.

And I think the Iron Man analogy is still really great here. Tony Stark's Iron Man suit augments him, right? He can do all these amazing things, but could also fly around on command, could do some basic tasks without Tony. And my experience with these companies has been that human tolerance for failure or hallucinations or lack of reliability, it just reduces dramatically as latency increases, right?

So the path of least frustration today for many domains is to build great augmentation and then just ride the wave of capability because we know it's coming. And so my advice for many domains would think about, like, build the suit and you can extend out to the suit that flies on its own once Tony or any of us is wearing it.

I'm not going to go through each of these, mostly because I lost time, but there are a ton of opportunities. We put requests for startups on our website. We're interested in a couple different categories of things. They go from, like, just good fit for purpose, like the law is a space of lots of text generation, right?

To things that weren't possible before AI. My partner, Mike, will say, like, this is a really interesting era of machines interrogating humans. What can you do if you can go, like, collect data on demand from people? We could talk to every customer, not just the top 5% by contract value.

We could root cause every alert proactively, right? Versus, like, just firefight. And the mental model is how can you build as if you had an army of compliant, infinitely patient knowledge workers. You know, one aside here is I think there are many hard problems where, like, the basic premise is the answer to them is not in common crawl, right?

The reasoning around them is not in common crawl. So this would be robotics, biology, material science, physics simulation. They require clever data collection, probably interaction with atoms, not just bits. Super scary for a software person, but I think the juice is worth the squeeze, right? The same reasoning that crushes math olympiads can seemingly navigate molecular space, and I think there are some really fundamental questions for human society that can be answered when people work on these problems.

And it's really cool as a machine learning person to meet people in their, at the top of their field at the intersection of machine learning in all of these other areas because, like, you guys would also understand, the same architectures apply, right? And that's just, that's really exciting. How should we think about defensibility to this advance?

Okay. So one last point, and then I'll conclude here. Some would say stay out of the weight of the labs, don't pick up pennies in front of the steamroller, right? But I would offer what I think is an uncomfortable truth. Execution is the moat in AI, and that's available to all of us.

Cursor arguably did not invent code completion, they did not invent the model, they didn't invent their product surface area, right? They just out-executed on every dimension of this. They shipped a great experience faster than their competitors could copy, and they captured the hearts and minds of developers, at least in this term.

I don't mean this to be cruel, but I often get asked about, like, counter cases and the importance of first-mover advantage. Let's be brutally honest. In contrast, like, Jasper had first-mover advantage brand that raised $125 million, but its first product was a series of prompts in a text box and, like, very good SEO, and, like, you have to keep running.

Like, ChatGPT, you know, crushed the first iteration pretty quickly. And so I don't think this is satisfying advice, but I think it is, like, real from the trenches. Build something thick and stay ahead, and, like, no domains are out of question. Magical AI experiences, they build customer trust and drive adoption.

And a lot of the data we need to improve these experiences and the context we need, it is not easily available today. And that advantage is, you know, open for the taking and not for the labs. So I guess in conclusion, I think the opportunity is early and really massive.

Like, I've made a career bet on it. I think many of you are. We're in the dial-up era of AI, and we're moving pretty quickly to broadband. Instagram came four years after the iPhone. Like, I was there when Greylock made that investment. Uber, five years. DoorDash, six, right? So the truly transformative companies, they weren't necessarily the first people to recognize the changes or the opportunity.

It was those who reimagined the experiences. And the game board keeps getting shaken up. That's the thing that's different this time, right? It's like getting a new iPhone. That's actually different every 12 months. And so you have, like, new model release, new capability breakthrough, you know, one-tenth the cost.

And every time the game board turns, I think there are, like, there's an opportunity to to win again. Okay, so I'll give you one last sentence and be chased off the stage. This was not my fault. Here's what I really want you to remember. You, as the engineers, got the magic first.

The anthropic, like, economic index says that 40% of use was still coding. That's not, like, 40% of the economic opportunity in the world, right? And so it is the job of everyone in this room and, you know, globally online to be the translators for the rest of the world.

So I encourage you to build something revolutionary. Thanks.