The pace of artificial intelligence research has exploded. Every week, thousands of new machine learning papers are published. But here’s the real problem: no human team can keep up, test everything, and still build something new.
That gap is exactly what Autoscience is trying to solve.
The San Mateo-based startup has raised $14 million in seed funding to build what it calls a fully automated AI research lab. The round was led by General Catalyst, with backing from Toyota Ventures, Perplexity Fund, MaC Ventures, and S32.
The fresh funding will be used to expand Autoscience’s platform and bring it to a select group of large enterprises, including Fortune 500 companies.
The startup also plans to grow its engineering team as it pushes further into automated AI research.
Building an AI lab without humans
Autoscience is taking a bold approach. Instead of hiring more researchers, it is building AI systems that act like researchers.
“We’ve reached a point where human intuition is no longer enough to navigate the complexity of algorithmic discovery,” said Eliot Cowan, CEO of Autoscience. “We’ve built a research organisation where the researchers are AI systems. We aim to compress a decade of machine learning research into months, unlocking new AI capabilities for scientists and forming a competitive edge for our customers.”
The company has developed a virtual lab powered by “AI scientists” and “AI engineers.”
These systems can generate new ideas, test them, and turn successful ones into real-world machine learning models.
In simple terms, it is trying to automate the entire research cycle. This matters because the bottleneck in AI is no longer data or computing power. It is human capacity. Teams simply cannot test ideas fast enough anymore.
Autoscience’s system splits the work into two parts: One AI system focuses on generating and testing new algorithm ideas. Another focuses on refining and deploying them. Together, they aim to replicate what a full research team would do, but at much higher speed.
The startup has already shown early signs of what this approach can do.
Its autonomous system became one of the first to produce a peer-reviewed research paper at an ICLR 2025 workshop. It also secured a silver medal in a Kaggle competition, competing against more than 3,300 human teams.
That marks a shift. AI is no longer just assisting researchers. It is starting to compete with them.
Targeting high-stakes industries
The company is already focusing on real-world use cases where better models can directly impact business outcomes.
Its early deployments are aimed at financial services, manufacturing, and fraud detection. These are areas where even small improvements in models can lead to major gains.
“We believe Autoscience is tackling an increasingly important challenge in machine learning: the pace and scalability of experimentation,” said Yuri Sagalov, Managing Director at General Catalyst. “As research output continues to grow, teams are looking for ways to more efficiently test, validate, and translate new ideas into production systems. We’re excited about their progress in advancing autonomous R&D to scale that workflow.”
