The AI boom is creating extraordinary opportunity - and significant risk. This panel tackles the hardest question in the market: how to tell which AI companies will endure, and which won’t.
Leading investors share how they separate substance from hype amid inflated valuations, fast-follow products, and aggressive burn. The discussion focuses on repeatable signals: defensibility, data advantage, routes to revenue, and whether teams are building businesses with long-term consistency rather than short-term momentum.
This is a practical session for founders building in AI and investors allocating capital in the space - grounded in judgement calls that repeat across cycles, not trend-driven optimism.
If you’re building, backing, or evaluating AI companies, this conversation offers a clear lens for identifying real value.
Auto-generated transcript - may contain errors.
Tap a timestamp to jump the video.
The, title of this event promises a bloodbath. We'll do our best to deliver. There is a lot of controversy up here. But, I work for Sifted, which is a media company covering the European startup scene. And in our daily newsletter, we report on the most recent funding rounds, the ones that have happened in the last twenty four hours.
And as one of the people curating that list, I am overwhelmed by the number of AI companies that we report on basically every day. I also feel that a lot of the startups that we're reporting on, seemingly do the exact same thing in the same way.
So it's no surprise that the market is starting to feel very saturated and, perhaps a correction or bloodbath is, you know, inevitable. So I was hoping for this conversation today that we can all talk about how that might shake out and what investors investors can do to better prepare themselves and and fine tune their portfolio as they're choosing amongst the many AIs companies that are coming up.
And from the other side of that, if you are building an AI company, how can you put your best foot forward? How could you create some long term value that, you know, these investors are looking for? As they say, start getting more particular.
And so we've got three amazing early stage investors on stage here. So I'm gonna let them unpack that. I'll first open with a question of what do you think is most overhyped in the AI field today? Gil, you've got some big opinions. Maybe you can jump in.
Well, look. If if you don't mind, I'll I'll I'll I I don't know if I can pick one thing that's overhyped or one broad area that's overhyped. I think, you know, there's there's sort of one or two ideas that I'll put on the table to start with.
One is, I think what we're witnessing in AI is a psychological moment as much as we're witnessing a technical tech technological moment. I think, obviously, there was that chat chat GPT moment. LLMs are a real thing. You can do a lot of interesting things with LLMs both on the application layer and also, I think, in an underappreciated way on the infrastructure level, meaning what you can do with them as part of a broader search and retrieval, you know, aspect of application.
But I think the the main thing that we're witnessing is something that was articulated to me by by a a technologist that I work with, who said, you know, shortly after this chat GPT thing happened, he sort of said, you know, what's remarkable about this is that everybody has had the same realization at exactly the same time.
And so you saw this whole wave of people building all the same things at exactly the same time. And that is, I think, particularly dangerous and it's kinda unique because as VCs, we're sort of in the business of spotting things early when there's only one person in some laboratory somewhere or deep in some, you know, development team buried somewhere within Microsoft who has some insight or some idea.
And and that wasn't happening at all here. This was everybody with with any desire to start something suddenly looked at what ChatGPT could do, and they they read all the, you know, expectation what it will be able to in the future, and they started startups at exactly the same time doing doing exactly the same thing.
That's super, super dangerous. So I think the the the sort of the the collective and then, by the way, the customers were the same. So the customers all you know, every Fortune five hundred CEO went home for Thanksgiving in twenty twenty three, I think it was, met their grandkids, saw ChatGPT, and then went home and told the board, hey, we have to do AI.
And then everyone started buying stuff. And I can get into more details on on on that. But that's what we saw throughout twenty twenty four and twenty twenty five, the demand was just exploding. And so as a VC, the other thing we see is revenue is sort of illusory.
It's not it's not as durable as we think it should be because a lot of these things are not getting sold. They're getting bought by customers with crazy demands for stuff. Some of that's great, but it doesn't necessarily mean that, you know, revenue means that you have an interesting business.
I'll I'll stop here and Okay. And Matt, what's your heart take? I think that media is probably the overhyped aspect of it. I think we live in a world where the media believes that we that we want drama, that we feed off that somehow, and that leads to revenue.
Mean, talking to a journalist. Think we're all getting well, I certainly feel pretty tired of that. The hype factor, signals to noise. And so I think that, to Gil's point, when the grandkids showed the grandpa chat GPT, they went back and created their own hype, and the media built on that.
We live in this echo chamber where people think we need to shout loudly about everything, and that's what's happened to AI. As VCs, our job is to try to reduce that noise, to try to find spots where we will invest. I think the depressing conclusion of what potentially Gil's saying is how do you invest?
Is that the blood path we're going to spend some time talking about? How do you invest in this sector? That's our job. I think there's two aspects of that. One is how do we find those niche areas, those unique parts of the AI landscape that we want to deploy capital into?
And the second thing is how does it disrupt our own jobs? I mean, maybe that's not what we're getting into today. But I think we all probably in this room feel some aspect of that is how is my job going to be different next year from this year because of this AI thing.
Right. And Denise, you know, your your VC focuses on media and tech in that space. So you have a bit more of a, like, a focused lens on this issue. So when you're thinking about what's overhyped in AI, how does this impact your your space?
I mean, similar to the other speakers, we are obviously seeing such high volume of very similar propositions. In particular, I guess when it comes to AI assistance or consumer facing applications, not just B2B applications, but also anything that has to do with consumer where it's incredibly difficult to have differentiation or any sort of credible user acquisition strategy or any credible business model.
And again, it goes, I think another aspect of why there's perhaps such a flurry of unstable innovation is that it's cool tech, it isn't necessarily solving novel types of problems or problems that there's a customer for, a client for, that there's budget for, but it's just kind of you know, exciting and cool, which is understandable, but not necessarily investable or interesting I'd say in the consumer space, there's definitely lot of hype.
I think agents, or the word agent is abused significantly at the moment. It's almost, you know, now everyone's building agents not software. I think it's kind of, and if we all agree that that's what we mean by agents, then maybe that's fine. But kind of walking into a pitch and sort of trying to borrow some of the terminology is just a bit distracting.
So, I do think that there's a lot of, of course there's hype. I mean, on the other hand, it's a good thing. I think it shows appetite to build cool stuff. And out of that, there will be some paradigm shifts. I think part of the problem at the moment is that, you know, historically I guess when you've had innovation cycles, you've had either new platforms emerging and then sort of the next generation is new applications being built on top of these platforms.
And then perhaps these platforms then evolve in the next cycle. And then there's like a wave of new applications. Whereas now, it's kind of all changing simultaneously. So, the platforms and the stuff being built on top of it are kind of all changing at the same time.
And also, I think the other issue is that generative breakthrough. I mean, there was a kind of moment where there was a breakthrough. But actually, the technology leading up to that has been a long time coming. And the appetite for generative AI has been a long time coming.
And so people have had ideas for what they'd wanna build. You know, were that technology available to them. Which is why I think that's adding kind of some extra, you know, sort of, kind of extra activity perhaps now or in the last couple of years that it's been a thing.
Anyway, but yeah, I mean, definitely busy. Definitely, it's clearing up. We're all in agreement. Too many AI companies in some sense. A lot of ideas being brought out. But how do you find a quality AI company through all you're sifting through all that.
You got your VC hats on. You're looking for an early stage AI cut to invest in. You know, what are some signals that a company could send that would stand out from the crowd? I'll leave Matt if you're nodding along. I'm staying too close.
I think distribution probably seems like a key advantage when we're looking at these things. Have they got a lock on some unique or or scarce form of distribution? Have they an example? Sure. Have they identified a way to reach a overlooked or not served customer base?
Mean, we're going to stop calling it AI at some point, but with their automated system. An example would be a lot of the consumption of AIs with tech companies, large, medium and small. What about the ninety percent of companies out there that are not technology companies?
Your plumbing companies, whatever it might be. And do they get access to these magical tools? In many respects, is magical. And so, that's a customer base that is broadly, very broadly overlooked by these companies and certainly by venture capital. Okay. Gil? I wanna get into the bloodbath thing a bit more.
So I I think, you know, what why do we talk about bloodbath? Right? I mean, I didn't write the title of the panel. I love the title of panel. So, you know, I think what is that? How does that unfold? I think part of it is if you looked at sort of going out of coming out of ZERP, the whole tech market realized that we had invest I I mean, I I think there are growth investors that were that sort of looked at their books in early twenty four,
late twenty three, and suddenly realized every single deal they've done over the past three years was a was mispriced. Right? There's very few companies that are at a hundred million dollars of revenue and growing, you know, really, really nicely and can justify that, you know, multibillion dollar valuation.
We've seen it in the public markets time and time again. These companies meet reality. The companies meet any kind of market turbulence. All of a sudden, their growth rate drops from a hundred percent to eighteen percent or something like that. And then, know, okay.
You're a software company with high margins. You're still high gross margins, but you're still losing money. You're you're doing over a hundred million dollars a year of revenue. You're growing at twenty percent. What is that worth? And who wants to finance that? And I think that was the initial SaaS bloodbath.
And the thing is the venture capital world and the tech world, like, these things often have you know, these companies have been financed. They have a lot of cash on the books. They can they can survive for a long time. There's no there isn't gonna be a single cataclysmic event where everyone suddenly goes bankrupt at the same time.
Right? But people are gonna look back at those vintages of of venture investing, those vintages of startups. And by the way, that whole company, that whole team, that whole CEO and founding team that could have sold the business for five hundred million, and they all would have been they all would have done great, and said they raised it a billion, they raised it two billion, and now they're screwed.
Right? So that's the kind of that's the bloodbath, but it's a silent bloodbath. And I think what's happening with AI is AI gave and and the VCs, by the way, in response to Zurp, raised enormous funds, enormous funds. And there's a there's plenty of blog posts out there.
There's one recently called the venture arrogance score, which is like, what percent of total exit volume in the world does your fund need to get to justify its returns. Right? And you realize some of these funds, like Andreessen Andreessen Horowitz's new fund, needs to needs to have a hundred and fifty percent of total venture exits over the next three years for them to make the returns that they're promising on their twenty billion fund.
It's impossible. Right? So the math is totally broken. And then AI suddenly showed up, and everyone's sort of willing to suspend disbelief again because there's magical new technology that justifies crazy fund sizes, justifies crazy valuations, allows people to shovel a lot of money really quickly.
And I think that's what leads to the sort of continuation of the bloodbath and through the or the the bloodletting adventure where this social technological phenomenon called AI, which is super real, by the way. I'm not like an AI denier. I just don't think there's good investments. Right?
So people are gonna invest at whatever price and whatever crazy thing and and and justify huge funds and huge rounds, and that will eventually play out in a negative way in ninety nine percent of cases. How what kind of timeline do you think on that?
Because right now, I feel like we're in the thick of it. Valuations are still sky high. We got investors knocking on doors, trying to get in on next rounds. It doesn't seem to be cooling down anytime soon, which is a great thing for AI companies.
You'll never notice it cooling off. It's gonna happen, you know, as as these funds mature, these companies mature. I mean, you see some spectacular things like eleven x. You know, there's stories like that where where sudden someone rips the covers off of a company.
There's other company even even Cursor, someone who knows the internals of Cursor or is in a position to know the internals of Cursor, is an amazing company, amazing product. Everybody loves it. It's it's probably one of best companies ever. As far as I understand, they're still losing money on a because they're they're they're basically selling tokens for less than they're worth to grow.
So who wouldn't buy that? But you can do most of what you can do with cursor with standard, you know, clawed prompts directly. You don't actually need cursor. So there's you know, they built a machine that works. I'm sure it's gonna be a great company, but there's there's lots of questions even on the highest flyers, and then there's the whole long tail of the ninety nine percent of companies Do do you really think it is gonna be a great company?
I don't know. I unless you show me their their their tech stack and you show me their their gross margins and you show me what they're paying Claude, I have no idea. It just seems a really good example of kind of companies we're talking about where it's so easy to copy.
Yep. It comes out, it ramps quickly, I'm sure it ramps. I don't know the revenue numbers. Then it's just very easy to replicate. So where's the unique long term defensible value in a company like that? And also the other thing I've been hearing about these companies is you kind of, and maybe this is another aspect of the debate is what do the humans do in this situation?
And actually a lot of Genie developers using Cursor, but they don't understand the code that's being produced. And so getting into customer issues and the user experience isn't what it needs to be. Again, it's all very early, this is anecdotal. But is that a great company?
It's automated a chunk of what we sent everybody to university to learn how to do. But you must become developers and your kids must become developers. And now, sort of, the media says, maybe not. But yeah, I don't know if that's a great company.
We've got examples in the portfolio. So you made a point about revenue, strength of revenue, durability of revenue. Companies in portfolio that grow quicker than I've seen a company grow in terms of revenue. They've gone from zero to five million dollars of ARR in no time at all.
Is that a good company? Don't know anymore. I think retention and churn becomes incredibly important in that. But we haven't had long enough to see if those cohorts are real or not. So without the right benchmarks to judge an AI company, you find it really difficult to analyze them relative to other things in your portfolio?
Yeah. Look. I think as a seed investor, if you're if you're investing really early in founding stages when there's no revenue and no traction, and and if you're in a world where the winners are gonna be characterized by sort of this ten x number that Matt's talking about.
Right? These crazy growth rates that no no I've never seen before. Right? I don't I am not equipped to predict that. I don't know how to predict that, especially in a world where the tech is so fundamentally deflationary. Right? The tech is so powerful that doesn't make your company better.
It makes all companies better. So so it makes your company worse on a relative basis because all your competitors have access to the same super powerful tools that are changing really fast. So that changes what we're we're looking for or it or it refines what we're looking for.
We're no longer looking for or or we're less interested. We're even less interested in, you know, rapid revenue growth in the early days or or some kind of thesis on why we'd get that. We're more looking for, okay, when the whole world is aware of what we're doing and the whole world can try to replicate what we're doing, is there anything at all that we have that's unique or sticky?
At the very least, sticky and ideally unique That would make it hard for someone to switch off of our product. Right. Because if it's not unique, we could debate that. Who's going to give value for it? And I think that is also part of the bloodbath is that there were no exits.
And so, with these like typical AI companies, the big tech acquirers, the big M and A producers of liquidity for our investors and therefore for our funds and for our subsequent funds, they believe they can do all that stuff themselves. So they don't need to buy these companies.
In house builds are now very big within especially bigger corporations. They'd rather do it with an on their own data stack as well than get external. So it's exactly that point. I think a lot of these players just aren't going to be able to last these kind of big shifts in the market.
But, I just want to pass the mic over to Diesel a little bit to get her in there. But when you're looking for AI companies, you know, what stands out to you as a defensibility for them? Yeah. I mean, from my perspective, I guess it's important to maybe go back to first principles.
So, just to agree with what you guys have said, know, is there really a customer that wants to buy from you that has budget and that will prefer this sort of external product instead of trying to build in house? A very easily forgotten question.
It's a very basic question, I would say, for someone to ask though, and perhaps more important. I mean, my perspective, ability to commercialize and generate revenue is actually important. I agree that the quality of the revenue has to be scrutinized. And that revenue isn't necessarily reliable.
But I think we shouldn't forget that up until a couple of years ago, early stage VCs didn't necessarily care about revenue or path to profitability. Whereas actually now, I think it's a lot more important and I think that's correct. And it's not just the kind of AI unit economics and margins that have to work, but also given the macro.
You know, there's a sort of an economic situation at a macro level that's happening in parallel or sort of independently of the AI hype. The reality of the markets that companies operate mean that having healthy unit economics despite AI technology where cost structures are opaque is important.
So, would actually scrutinize that heavily. Of course, team. You know, VC will always say we invest in people first. And that's more true than ever, right? So, ability to execute and to deliver and distribute and figure out your channels. Hiring is probably another area of, you know, again, thinking about team growth.
So, the main challenge for CEOs is building the right teams around them. And I think that's more important now given new types of engineering challenges that developers have historically not had to deal with. Haven't had years of skill set building in. Working with GPUs is quite a novel thing for many developers, for example.
So, how do you think about talent and extra talent costs? So, I don't know that, I guess, what we look for is novel. But perhaps the challenges around or kind of how we think about each of those domains may have adapted a little bit in this kind of new AI reality.
Well, so anyone building an AI company. You know, if they're starting now or they're still kind of shaping it, you know, what are some things you'd like them to keep in mind so that they could build a strong company that, you know, could be on the radar of you guys?
Any kind of I'd say the customer. I feel like, don't say this for the third time. Just, you know, be clear on what problem you're solving, who's your customer, and that what you're building is something that your customer wants. You know, think we forget about this very basic, I guess, mantra when it comes to lean startup, right?
At the end of the day, we're trying to build products for customers. It has to go back to them. Solving a real problem for customer. Oops. Right. Yeah. Think some of the basic things like Denise is saying, I always like to know how many customers or potential customers have you spoken to.
And it's amazing like how how few people come at this idea, for their view on the world, product has to exist. I'm going build it. I'm full of ambition to do it. But they haven't kind of proven to themselves yet by doing the data, by going out and speaking to customers or potential customers, would you buy this product, etcetera.
It's very basic, but I mean, that's just something I would look for. Fair enough. I think for us, it probably comes down to, you know, we're looking for some measure of depth. It's either, you know, I I hate the phrase, but deep tech.
You know, it's either there's some, you know, for example, we backed the graph database company that, you know, the guys have spent their whole careers building graph databases, and their graph database is a hundred x faster than NEOs. That expertise is a lot rarer and harder to find than the kind of vector LLM expertise that exists today.
So some kind of deep tech something or some kind of deep domain expertise where yeah. Yeah. You can vibe code all you want, but you can't vibe code your way to a vertical solution for a vertical buyer that doesn't wanna buy software at all.
Right? That's a really hard thing to do. So you have to build a full stack company that's not just we have some code that does some stuff. We have a whole way of going to market and delivering value to customers. You know, that the company I'm thinking of is a company that sells software for the shipping industry.
And they got started before the AI boom, and now they're they're rolling AI into the product. It it doesn't matter. It's like and anyone can roll AI into a product. That's not hard. But selling you know, they're deployed at a hundred fifty terminals around the world, that's hard to do because those people don't wanna buy software.
One of the frameworks that we use, if this is helpful to people building, thinking about raising venture capital, is we think about it as a a novel system. And that speaks to a lot of things that Gil's talking about around this horrible deep tech, But like, what is your true stack?
What is the intellectual property in there, whether it's hard or soft? What is the team's right to win in that space they're building in? Do they have viewpoints about the future of the way the industry is going and they're building technology into that unique and rare viewpoint?
We think about that as a system. What is your novel system that you're building as opposed to what's the software? Just the software. It's a combination of the team, the market dynamics, and the tech and product that you're building. And all of that is encased in this idea of a system.
Great. We've got about seven minutes left, and I was told that we're able to ask questions from the audience because we've got a spare mic floating around. Fred's definitely trying to get his head up here. Hi, guys. I'd love to hear an example of a company you've backed that you love.
That's a neighbor native AI company that's killing it and what they do. So maybe flip it around. And Fred, what what do you mean native AI. What's your definition of native AI? You know what? Pick whatever AI company you like. I'm happy to give a definition, but I'm narrowing the field.
And maybe Denise, know, go for it. Sure. Yes. So we we've backed a company called Electric Twin that uses synthetic human behavior simulations for the purposes of market research and product development. And so, I would say it's through and through AI native in the sense that the type of, well, the type of queries that you can do, but also the type of product that you can build is just, you know, wouldn't have been imaginable beforehand, pre generative AI.
They're also, I think, being extremely scientifically rigorous with how they're doing this. And of building in the right way in terms of guardrails for AI, responsible AI building, and really enabling the type of research that's privacy first and compliant as well that I guess would previously have been more challenging as well.
So new product, AA native, as well as with all the protections and security in place that would otherwise have been quite challenging to enforce. We invested in a company in twenty twenty. Which is a generative AI company back then. And it's having its moment now, it's called Gizmo.
And it's a learning technology. So you can put in, I want to learn about the history of AI or anything you want. And you can feed it PDFs or any kind of content. And it will generate flashcards specific to you, what you want to learn.
And it will test you and it's gamified and the streaks, etcetera. And we invested four or five years ago before we were really talking about generative AI. And founders were just, they were deeply technical founders. And so they were building in a way that was accessing the latest tools and technologies.
And today it's suddenly having its moment in the sun. Not because of generative AI suddenly becoming hot to the kind of point of this conversation. But I think that it is plugging into the way people actually want to absorb content and they want to learn in this example.
It's in a tech company. So generative AI before we're talking about generative AI. Gizmo dot AI, if anyone wants to play with it. I'm trying to you know, your question was really specific. You said, you know, killing it and AI native, which I took to mean, you know, companies that were sort of founded post twenty three and killing it.
I take to me, like, having revenue growth that's explosive. I I we don't have any of those. And I think that's interesting. I think the reason we don't have any of is because we didn't we sort of almost deliberately didn't look for those.
And it was in twenty twenty four, there was this massive wave of sort of AI enabled application companies using generative AI to do and and a ton of them had had done reasonably well when we met them. They had revenue that we could have only dreamt of seeing in a company that hadn't raised money.
Right? So so if we'd invested in some of them, I could probably be sitting here telling you that we're killing it because, hey, when we invested, they had five hundred million sorry, five hundred k of ARR. Now they have three or whatever it is.
But they were they were all, like, and I'm I'm sure you know what I mean by this. They're all sort of wrappers on top of LLMs. They were like, we know how to submit a regulatory filing really quickly or we can, you know, create a NDA really efficiently or some something like like that.
And we just didn't see any of those that we thought were gonna have a durable advantage. So we didn't do any of those. The stuff we've done that's been AI sort of since that period has really been infrastructural. It's it's related to what you said.
Our thesis is fundamentally that LLMs alone are not gonna get anyone very far, number one. Number two, that most of the value that's gonna get created and and and generated by this wave of technology, if you look at software vendors, is gonna be created by companies building their own internal tools with open source models, not the big foundational model players, and they're gonna be looking for tools to make that possible.
So we've done this graph database, which is a graph rag company, which enables enterprises to build their own rag solutions. We've done a company right now that does model distillation in a really interesting way that allows you to build really, really efficient and cheap models that are that are really cheap to run, and you can run them on a on your own proprietary hardware.
One of their potential customers is is a security company that knows that every single one of their customers is gonna need their own generative LLM style models running on sensitive security related data on their premises, and that needs to be done very cheaply and efficiently.
Killing it, none of them are killing it yet. They're really early, because, you know, we we just didn't we didn't buy into that thesis, I guess. We have probably question for time for one more question. Has one? I have a question about for all of your portfolio companies that you've invested in in the past that are now being asked to incorporate AI into their products.
And while you're simultaneously watching the startup landscape in AI rapidly developing products and tools, what kind of guidance are you giving your more established portfolio companies on what parts of the AI roadmap to bet on? And what's going to be commoditized in AI?
What's you know, actually value what's the differentiating AI strategy for a more established company today? I think we're we're not doing that. I mean, I think that there's all this hype about AI. But by and large, the companies that we've backed, it's kind of like what Gil's saying.
They're not we didn't back them because they had this amazing AI. Most of them are using it. Most of have been using it for a while. Most of them integrate it into their stack in, you know, in different ways. So it's really hard to give you a cool answer to that question because we just haven't invested in these pure, like this is an AI for this and AI for that company.
Companies that I really like are companies that have been doing it themselves in clunkier ways. The AI is allowing them to accelerate that part of their process and that part of their build or make that part of their a small part of their product that much better.
So, I've I've I've got two specific answers for you. I I like this question more than Fred's question because I have a good answer for this one. Two specific examples. One is both companies started before the sort of ChatTPT moment. Right? One is this software for shipping. Right?
One of the things they do among many, many others is alert ship owners when they should slow their ships down to cut emissions if the port's not ready for the ship to arrive. And and the the the the it's a four billion dollars a year potential savings, not to mention the carbon.
That but they didn't Gen AI allows them to write an email. It's super simple. Like, anyone can do this. This is not hard. Right? That's the easiest thing. Any developer can do this in a day. Right? So there's certain places where that is really easy for where Gen AI suddenly allows you to do something cool in your product that would have been actually really annoying to code up previously.
Right? So that's one example. A second example is, I think, more interesting. We have a company out of Israel that's building if you guys are familiar with Figma, it's like vector thing where you can draw interfaces and front ends. And then you give that to a developer and he writes code or, you know, she writes code.
And Figma has some some tools that can convert from vector to code, but it's literally impossible to go back from code to vector. So once the designers give it to to the developers, it's game over, then we have to have meetings every time we wanna change a color.
These guys had had been banging their heads against the wall for two years trying to build a bidirectional translation engine so that you could have a single tool to do code and design. So that a developer could write code and a and a designer could actually draw vectors around and and wherever, and they could talk to each other without having to move tools.
It turns out that's incredibly, incredibly hard. Long story short, these coding models appear. They fire seventy five percent of their team, and they've now built this themselves, and it it works. Because all of the pieces that were missing, which were impossible to code in a discrete way, you can code with LLMs.
It the exact missing piece of the tech that they needed to make that whole thing work, and they just demoed it to me this morning and it works. So so there's there's stuff like that where now that's only interesting because they built this whole substrate that's really really hard and complicated, and you can't just copy Right?
Because when you go and and design that interface using a prompt and they spit the interface out and they can then render that in code immediately, you can open up the editor and you can actually drag the vectors around and it works like Figma.
That's hard. That's you can't LLM that. Right? So there's stuff like that that's, you know, where you have a company that built, you know, ninety percent of something and then the LLM just happens to be that ten percent missing piece. That's really interesting stuff.
That's great. We're past time, but do have a fairly quick one or? I don't think I will bring a lot of insight. I'd say this is a real problem, a real risk for many mature businesses. I'd advise our portfolio companies to think about their customer and solidifying their value proposition.
So, it will depend on the case rather than generic kind of Well, you so much, Matt, Gil, Denise. Appreciate it so much. Thank you everyone for joining us. Thanks a lot.
