We raised a $16M Series A led by Standard Capital to build the software that turns data into better AI. Reinforcement learning is how frontier models improve, and HUD is the platform where anyone can build and sell the RL environments that make that happen.

Our round included participation from Y Combinator, Exceptional Capital, Liquid 2 Ventures, and Twenty Two Ventures, with angels investments from Dylan Patel from SemiAnalysis, Roon from OpenAI, Ivan Burazin from Daytona, and Theo Browne.

How We Got Here

We started HUD after building computer use agents at Hume, where we kept hitting the same wall. The agents were improving fast, but the infrastructure to train and evaluate them barely existed, so we decided to build it.

We built our own benchmarks and they got used right away. Autonomy-10 evaluated OpenAI's Operator at launch. SheetBench-50 was validated by PwC, Cisco, Charles Schwab, and Fannie Mae. OSWorld-Verified fixed more than 300 issues from the original benchmark.

From there we worked directly with leading model labs on the environments they needed to train against. We opened a vendor marketplace so independent builders could supply that data, and vendors on it are already doing millions a month.

The incumbents who had lab access kept it intentionally. Their moat was being one of four or five trusted vendors. We built HUD so that moat stops mattering.

What We've Built

HUD provides infrastructure for LLM reinforcement learning. Model training More than 50 businesses now run on the platform, and our vendor marketplace moves millions of dollars in training data each month.

Why We Built HUD

RL is the mechanism by which AI gets better. Every capability gain in a frontier model comes from running agents against real tasks, grading what they did, and training on the results. The data that feeds that loop is the most valuable input in AI development, and about 90% of what labs buy today never makes it into a model.

HUD enables any company to hill climb model improvement with your own data. Spinning up environments, setting up tasks, and measuring performance gain is the core loop HUD enables. Anyone with domain expertise, whether a finance professional, a DevOps engineer, or a legal researcher, can now build a real RL environment, run it against our public benchmarks to prove it works, and sell frontier-grade training data directly to the labs that need it.

Dalton, our lead investor from Standard Capital, put it simply: "HUD is to ScaleAI what Airbnb is to Hilton".

What's Next

We're scaling the vendor marketplace and adding new environments across coding, computer use, and agentic tasks every week. Our goal is to let anyone, anywhere build and sell frontier-grade RL environments and post-training data.

We're hiring engineers, researchers, and data entrepreneurs who want to build that future with us. Come see what we're working on at hud.ai.

We sat down with Dalton from Standard Capital to talk through the raise and where this market is going. Watch the full conversation:

HUD + Standard Capital Series AWatch →