Autoscience Raises $14M in Seed Funding to Build Autonomous AI Research Systems
Autoscience just walked into the lab, turned off the lights, and let the machines cook. $14M in seed funding, led by General Catalyst with Toyota Ventures, Perplexity Fund, MaC Ventures, and S32 riding shotgun. Not bad for a company whose whole premise is that the next great researcher might not need coffee breaks or conference badges.
Credit where it’s due, Eliot Cowan, CEO, didn’t build another AI tool that politely assists. He built a system that replaces the entire late-night research grind with autonomous AI Scientists and Engineers that don’t tap out. These systems generate hypotheses, test them, and push validated models straight into production. No hand-holding, no “circle back next quarter.” Just output. The kind enterprises actually pay attention to when models are tied to revenue, risk, and reality.
And before anyone rolls their eyes and says “sure, another AI claim,” the receipts are sitting right there. A peer-reviewed paper accepted at an ICLR 2025 workshop. A Silver Medal in the Kaggle Santa 2025 competition against 3,300 teams. Not theory. Not vibes. Performance, measured in public, where the scoreboard doesn’t care about your pitch deck.
The real play here isn’t just automation. It’s compression. Machine learning research has been bottlenecked by human bandwidth for years. Thousands of papers a week, and only a fraction ever make it into production. Autoscience doesn’t read the backlog, it eats it. Hundreds of parallel AI researchers running experiments while most teams are still scheduling their next sprint planning session.
That’s why the focus on high-stakes environments matters. Financial models, manufacturing systems, fraud detection. Places where a 1 percent improvement isn’t a vanity metric, it’s margin, efficiency, or millions saved. Autoscience is positioning itself as the quiet force behind those gains, the engine under the hood that keeps getting smarter while nobody’s watching the RPMs climb.
There’s also a subtle shift in how enterprises think about building. Instead of hiring bigger teams to chase incremental gains, they’re starting to ask a different question. What if the research team scales like software instead of headcount? What if experimentation never sleeps? That’s the door Autoscience just kicked open, and $14M says some very sharp investors think it leads somewhere real.
