Basis Set Ventures closes $250M Fund IV in a single day, betting on generalized AI agents and physical world automation

handle your customer support, go to bin.ai. And now we will begin our Lambda lightning round. We have uh Victor joining us right now from Synthesia. He's the co-ounder. I

know we have we have

Am I out of Oh, I'm out of order.

Basis set Ventures coming on to announce a new fund. Let's bring her in.

Welcome to the show to the TV van Ultradom. How are you doing? Good. How are you? Thank you for having me.

Thanks so much for coming on. Uh

we have an incredibly large gong here. We're going to get straight into it. You have some news. We're ready for it.

John jumped jump to the fun part. But uh anyways, uh great to meet you. Great to have you on. Why don't you give kind of a a quick intro on yourself and the fun and then we can get into the to the new entity.

Great. Uh, I'm L. I'm the founder of Basis Adventures. We were one of the earliest AI focused funds before it was cool. And we just raised our fourth funds, which is a $250 million uh vehicle. Um,

thank you. I'm here just for this. This is the best sound effect.

Uh, what take us back to the beginning. When did you first start, you know, thinking AI is important, investing in AI, building in AI, like uh what what's the journey to get here? It seems so obvious in 2026 that it's a good investment category, but this is fund four. Uh what was the journey like to get here?

Yeah, I was actually a researcher before. I studied human brain functions which is this part frontal lovaria

and taught computer how to acquire language as children. This was the earliest of AI which was decades ago. Yeah. And uh I was uh fortunate enough to join Drawbox and really understand how

sure

not AI but how capital works. I grew up very poor.

So the education for me is more on the money side and venture side.

AI has always been super uh obvious uh to me. I started funds in 2017. were one of the earliest AI funds and I was actually fortunate enough to made an angel investment before I started a fund called Costco AI. Uh it was one of it was a was it was a good uh very good first investment uh personally and started uh bases in 2017 and

we made some pretty early investments uh before they became very obvious anything from uh AI infrastructure uh supply chain automation

uh you know such as Quins and I think Mezzero actually came on the show to announce their funding as well.

Yeah. Yeah. So it was it was great segments uh they uh you know mezzero and other companies make the system more intelligent because the current AI system have deficits right they are not very much like humans yet. Uh a lot of these infrastructure applications make our system more intelligent and we must seeing infrastructure and applications of these systems.

Uh why did your why did your fund raise only take a day? I was about to ask the same thing

I think because we we've been actually western broke the news today and the incred LP who uh who's been early LP and she said uh we've been very consistent disciplines and really stick to our uh kind of a story since day one we really have it for the eight past eight years we've always believed AI believed in the founders before it was obvious uh we really proud of oursel to be the first believer of the founders and market before the narrative uh catch up before the signals are clear.

So I think uh LP LPS can see that and uh they were very fast in committee.

That's good.

What's been your reaction to Cloudbot?

Oh. Oh man, I I love it. It's very empowering. At the same time, I'm a little bit scared of the security issues. Mhm.

We we actually have our internal ad part

just give your whole just say clubbot run my fund as a way to test it see how it does just deploys

we actually

deploys internally

that will make our life so much easier huh um we actually have internal AI partner uh we actually one of the earliest form to have a engineering team help my team engineers we run our internal systems we actually a lot of it is uh you know autonomous

uh but like we really care about security So cloudbot is uh you know why I really really love it as someone who's very optimistic about it and our future uh of intelligent system. It scares me a little bit uh from that point of view but it will get better. I think it will get better very quickly. Uh so everyone can you know chat with your phone and have your work done for you. How do you think about Oh, sorry. How do you think about the, you know, the the mega scale AI labs, anthropic and open AI versus the neolabs that are training new models versus the application layer AI companies? What's most interesting to you? What have you spent most time with?

So, I think all are interesting for different reasons. Like the way I think about it is today's models have deficits, right? They're not very much humanlike. For humans, when we do something, we kind of form a hypothesis, we go collect data and we test the data, we form actions, we we have memories, we come back, we revise our actions.

Today, uh the models are not there yet.

Lots of components are missing and it cannot form this loop. So either your foundational lab that trains the models to be more complete and can actually have a better world work view or you can be components that plug into the technology that make the current model better or you can be application in certain uh vertical such as you know supply chain construction and collect uh even more detailed data. I think all of them can make uh our future more intelligent and you know we're bullish on all of them but it really depends on the founder and depends on the company uh when we make investments. How do you think about transformation of legacy enterprises? Obviously, so many software companies are telling a story about AI transformation. There's a lot of startups that are working with enterprises to bring AI to bear inside those organizations. And then there and then there are a whole host of AI startups that are just saying, "Hey, we'll just do exactly what that big company does, uh, AI natively." Where's the bigger opportunity? I love transforming all kinds of the economy, not just anything on on your computer. I think our computer

We love We love transforming the economy, too. Let's give it up.

It's a great line.

No favorites. All parts, please.

Yes.

Well, our computer will be automated soon. I think within the next couple years, we have uh companies automating like computer use, right? You don't even need need to use your hands. The computer going to be automated. It's much harder to be a robotics company

uh working in physical world because our models don't understand force or temperature or a lot of the nuances that we take it for granted. So a lot of these data are much harder to collect and it's going to be uh taking much longer for us to automate like supply chain, robotics, manufacturing, real estate. Uh but a lot of these work is you know we don't have enough uh labor to do such work. welding for example. I think that you know we're we have a shortage of people want to be welders. today's young younger generation don't want to be welders even though you get paid a lot of money three to four $500,000 I think to be a welder but um but it's hard but it's hard it's hard work right so but machines not not ready enough to automate real work uh with force with you know temperature changes with uh you know a lot of hardware u which is I'm actually really excited that every parts of our the hard work that also will be uh automated so We as human can spend time doing things we actually like doing.

Like talking on the show.

That's true.

Good, good point. Um I want Yeah, I hate to go back to something we already talked about, but how how do you expect the like labs and other players to react to Claudebot? Like obviously people are using the product uh and they're excited about it and they're bringing up very valid concerns. And if you see something that's amazing or could be amazing and you have concerns about it, why not take another crack at it? Like do do you think this is something that um like we were debating earlier like thinking about like Google launching a product like this like they have the technical capabilities to do it but the kind of even legal uh overhead of of launching a product that theoretically is this powerful uh is immense. So how how are you thinking about just kind of agentic sort of like general purpose like agents and and kind of uh the as a category is with investing whether looking at existing portfolio companies or or new investments.

Uh I mean first of all if I were running large ad labs I go reach out and hire this guy immediately.

Yeah.

Yeah. I'm sure I'm sure you already got the the bigger the Zuck like okay $1 billion

or the seed round both are going to have seven zeros at least.

Exactly. Exactly.

Uh we're very bullish on this. I'm actually is a really great point. This is why one of the reasons we're really bullish on startups because you're able to do a lot of things that large labs are unable to do. Computer use agents for example generalized agents that's the future. Uh we have a company called similar that's all they do which is automate all the work on your computer and their generalized tool uh guaranteed to be better than the frontier labs because they're built on top of all the models and you're able to use any model at any time with this more generalized platform agnostic tool. So I'm very bullish on a future of generalized tool you know and also very specific tools as well. you have to build it's it's much more challenging to be generalized because your competition uh is is is much more right than you know being a welding company uh it's more specific uh so

what about uh what's your take on browsers specifically consumer browsers we were wondering earlier you know there's this excitement from various players new we you've had new browsers emerge we have the open AAI browser perplexity etc it feels like this new paradigm with agents maybe that is like kind of focused on the wrong layer of the stack potentially if people are just talking with their computer and then a computer is you know brow you know effectively using a browser on the user's behalf. Yeah, I mean to be honest, I think my hot take is browser. I don't want to automate a browser. I want to automate everything. I want a system level control. Just take over my computer. Don't take my browser. I have to still click on

Yeah. So, no, I completely agree. I completely

No. I think I think I think people's people's experience using like some of these like agentic browsers is like it feels like you're kind of like observing like like your grandma using using the browser, right? It's like maybe this isn't the right

paradigm.

Yeah. I mean ideally I just take my phone and I just chat with it and have 10 MacBook Mini that runs every and I don't have to do anything. That's

really the dream. So I think browser is a little bit limiting my view. I think we need system level automation computer use. I don't need to use my hands.

Well, it's going to be a fun time to deploy this fund.

What's your what's your sweet spot? Pre super early. You said preede where what's the latest stage that that uh you're investing all that stuff?

Uh we like to be as sort of early as possible. Uh we like you know our average check size about two to three initially. we go up to 10

uh initial check. We like two typos founders. If you're a founder who are working on really hard problems uh technology that probably don't even exist today, please reach out. Or if you're a founder working in the market competing with hundreds of competitors, but just faster in iterating and learning also please reach out. We like both. Obviously, they're very very different companies. Uh we're usually first believer before the market is too, you know, gets obvious gets big. So we'd like to be that first partner with you in the trenches and work very hard.

Amazing

for you.

Well, so great to meet you.

Yeah, congratulations on the fund more of your companies on and uh great great work on the fund raise.

Thanks so much.

Thanks so much.

Have a good rest of your week. We'll talk to you soon.