Investing in the 1–2 AI companies that matter each year
We built the platform used by 1M+ AI engineers at every major lab. We know what exceptional teams look like because we were one. Now we back the founders building the next generation of defining AI companies.
Our LPs
Improbability Engine is backed by the operators, founders, and institutions who built modern AI — giving our portfolio unmatched access to the people shaping what comes next.
Plus operators and builders from OpenAI, DeepMind, Google, Meta, and the teams defining the frontier of AI.
The Team
Managing Partner
Grew Weights & Biases from 100 users to over a million, making it one of the fastest-growing AI infrastructure companies ever built. A $2B exit.
I built the growth engine behind Weights & Biases — the product, the community, the go-to-market motion that made W&B indispensable to every major AI team on earth. From foundation model labs to Fortune 500 companies, I've seen and worked with every team building AI that matters. Deep network across every major AI lab and research community. I know what makes an incredible team because I was part of one. Now I back the founders building the next generation of defining AI companies.
Advisor
Founded Weights & Biases, one of the most important companies in the modern AI stack ($2B exit to CoreWeave).
Lukas has spent two decades building AI companies from research to production. At Weights & Biases, he created one of the only go-to-market motions that truly worked for an AI infrastructure company — product-led growth that scaled to more than a million ML engineers without a traditional sales team. The platform became core infrastructure for nearly every major AI lab and company. He is also the host of Gradient Dissent, one of the leading podcasts in AI, featuring conversations with figures like Jensen Huang, Daphne Koller, and Sebastian Thrun. Lukas brings rare experience across the entire lifecycle of AI companies — from the first ML hire to global infrastructure used by the world's most advanced teams.
Advisor
Cofounder of Weights & Biases and the architect of the ML infrastructure that scaled to billions of models trained.
Chris built the technical foundation that allowed Weights & Biases to become a core part of the modern AI development stack. The platform is used by nearly every major AI lab and company to train, track, and scale machine learning systems. With deep expertise in distributed systems, ML infrastructure, and developer tools, Chris knows what it takes to build infrastructure that engineers truly rely on. Before Weights & Biases, he cofounded Figure Eight, one of the earliest platforms for large-scale AI training data. He has spent his career building the systems that power modern machine learning.
For Founders
We write $250K–$2M checks at pre-seed and seed for AI-native companies. We move fast — term sheets within a week, same-week wires. No board seats. We've built and scaled an AI company from zero to a $2B exit — every problem you'll face, we've seen it and solved it. If you're building something that matters, we will be your fiercest believers.
We read every application carefully. If we think there's a match, you'll hear from us soon — often within a few days.
The Network
An invite-only community of researchers, engineers, and founders pushing AI forward. Builder dinners, hackathons, and the kind of rooms where the next great companies are born.
We'd love to learn more about you. We'll be in touch soon.