N of 1 raises $15M to build trading agents — betting that AI-powered investing will follow the same adoption curve as coding agents

Speaker 1: Mhmm. Well, congratulations on the progress. Thank you so much for taking the time to

Speaker 5: talk Great

Speaker 1: meet you finally. Great to meet you.

Speaker 5: Congrats to the team on the Thanks for having us.

Speaker 1: Have a good one.

Speaker 4: Take care.

Speaker 1: We'll talk to you soon. Up next, we have Jay from n of one. He is the founder. This is his first appearance. Building trading agents that may be the next AI interface. What's going on? What's going on, guys? Welcome to the show, Jay. How are doing?

Speaker 2: I'm great. How are you guys?

Speaker 1: Good. First time on the show. Please introduce yourself and the company a bit.

Speaker 2: Yeah. So my name is Jay. I'm the founder of NF1. We're an AI research lab focused on financial markets.

Speaker 1: Mhmm.

Speaker 2: So we're taking a big bet on on markets as the next environment that that leads to training models that are, you know, better at trading and better at domains that require like adaptive intelligence.

Speaker 1: What's your background? How'd you get into this?

Speaker 2: I have I've I've kind of got gone back and forth between public market investing and software building software companies. So it's kind of like bringing together the two sides of my experience.

Speaker 1: Sure. Do you think Renaissance Technologies secretly discovered the transformer back in the nineties? This is a conspiracy theory.

Speaker 2: I I don't I don't think so. We actually just hired someone from Renaissance. No. And, yeah, I I don't think so.

Speaker 1: It seems it seems unlikely given their their their compute bill over the years. I think we would know.

Speaker 2: Yeah. There be signs.

Speaker 1: There would be signs. There would be signs. But what is working in terms of financial performance in the AI world? There's two weird dichotomies, which is like a lot of the best and most profitable financial firms have been using machine learning for decades or a year

Speaker 8: or Yeah.

Speaker 5: Or the other side, you could say Jane Street is an AI lab.

Speaker 1: Yeah. And they've invested in custom hardware and computer infrastructure for a very long time. They wrote their own programming language. Okay. Like, they've been investing in this for a long time. And then on the on the flip side, you have the best frontier models when someone gives them a benchmark of, like, try and trade this portfolio. We haven't seen good results. So where is the the the delta in your opinion?

Speaker 2: Yeah. It it really depends on what your goals are. I think, you know, Jane Street is their goal is to make money, to make as much money per quarter or or year as possible. Our goal is to train models that can generalize across markets and eventually eclipse, you know, the best human traders and algorithmic trading systems in the world over the a very long time horizon. And

Speaker 1: for the fur

Speaker 2: but for the first part of our goal, it's to to make trading agents a thing, and that's mostly an infrastructure. They're honestly kind of like coding agents for markets. So we're building a consumer facing platform, not a hedge fund. And so our goals are are pretty different. We we we kind of see hedge funds as having a more short term goal of making money and and us having a much more long term goal of, you know, trying to use markets as a learning environment for AI in general

Speaker 3: Mhmm.

Speaker 2: Which leads to totally diff building a totally different type of company, hiring totally different types of people. Yeah. There's not many hedge funds that we know of that are trying to train models where their goal is generalization versus just, you know, making as much money as possible.

Speaker 5: I know. Talk about the consumer experience. Someone signs up, they deposit funds, your model start trading it. Is that is that roughly what you're aiming for?

Speaker 2: Yeah. So to take a a short step back, basically, our view is that trading agents are about to go through the same sort of adoption curve as coding agents. From basically no one using them today to basically everybody using them in the next couple of years. So like two or three years from now, we believe that trading without a trading agent will be like coding without a coding agent or a cursor

Speaker 6: Mhmm.

Speaker 2: Or feel like. You could do it, but you know, why why would you? And yeah. So that's like the sort of inflection that we see coming. It's like the next big platform shift in investing that we see. And, you know, you people have lots of ideas today around trading strategies or, you know, low resolution DCs they might have, where executing that is kind of hard. So whether it's a a systematic trading strategy or, like, some more complex discretionary one, you might have an idea, but, you know, it's just impractical to put it on unless you're good at coding and have a bunch of data and other things. So, yeah, basically, you'll be able to come onto this platform, you'll be able to describe a thesis or an idea in natural language, and then the models that we that we are developing will be able to get you from that natural language to a fully deployed trading agent that expresses or embodies your your viewpoint.

Speaker 1: I think there's someone in the chat who knows you. Do you know Moon from stable diffusion? You and Pedro kicked off his career? I don't know. He's he's excited that you're here. Anyway, I

Speaker 2: have no idea who that is. Sorry.

Speaker 1: Tell me about what the important pieces to build are for because you can, in theory, use OpenClaw or an AI agent to interact with, like, an interactive broker's API, sort of set up a trading strategy, work through that. How like, what are the pieces of the scaffold that you need to build to, like, give a great customer experience?

Speaker 2: Yeah. That's a great question. So there's a lot. The first hardest you know, the hardest part in many ways is the data. Mhmm. You need to have really good live data coming in from basically every market anyone wants to trade. So we spent a lot of time really hammering out all the different providers and cleaning and preparing and getting it ready to be consumed by agents. So that's a huge piece of it. Like, if you don't have all the necessary data, it's it's hard to to build a real trading strategy. The second piece is, like, the guardrails. You you know, LMs, like, a big a big part of our point and and why we launched Alpha Arena as as and maybe you guys saw Mhmm. Is to prove that LMs aren't quite there yet when it comes to trading. The the models aren't very good at at autonomous trading tasks, basically. So we developed like a harness and put a lot of our energy into to developing like a a world class harness for trading and deep research queries for markets. So you get all of that on our platform. And then all the DevOps and server management and basically all the infrastructure that you need to do that well, even on something like OpenClaw, we've take we take care of that as well. But, yeah, it's it's there's a lot. And then the execution, and then if you want to do paper trading or simulate out possible ideas or, you know, kind of, you know, go through a lot of iteration and refining your strategy. It's it's it's not it's not super easy to do to do all this stuff well. Yeah. That that's

Speaker 5: Would you ever would you ever partner with the Publix or the Robinhoods of the world and and integrate this into their platform where they can effectively create some type of tab where they can type you can type in a thesis and then execute it within your existing brokerage or do you wanna own the own the end customer?

Speaker 2: Yeah. As of right now, we're integrating with all the brokerages. So you'll be able to connect to whatever brokerage you use, or if you don't have a brokerage, you can just onboard through one that we set up for you like Alpaca or or MBKR. But yeah, we're we're right now, we're just a layer on top of all the brokerages.

Speaker 5: Got it. Cool. Well, thank you so much for coming. You have some news. You have some news. You raised some money.

Speaker 1: Oh, yeah. Oh, yeah.

Speaker 2: We just raised a $15,000,000 round. Thank