General Catalyst leads $14M seed for Autoscience
Autoscience, a San Mateo-based startup building what it describes as a fully automated AI research lab, has raised $14 million in seed funding as companies struggle to keep pace with a surge in machine learning research output.
The round was led by General Catalyst, with participation from Toyota Ventures, Perplexity Fund, MaC Ventures, and S32. The company said the capital will be used to expand its platform, grow its engineering team, and roll out deployments to a select group of large enterprises, including Fortune 500 firms.
Replacing the research cycle with “AI scientists”
Eliot Cowan, CEO of Autoscience, said the company is designing AI systems that function like a research organization rather than a set of tools that merely assist human researchers. “We’ve built a research organisation where the researchers are AI systems,” Cowan said, arguing that human intuition alone is increasingly insufficient for navigating the complexity of modern algorithm discovery.
The startup’s virtual lab is built around two roles: AI scientists that generate and test new algorithmic ideas, and AI engineers that refine promising approaches and turn them into deployable machine learning models. The goal is to automate the end-to-end research loop—ideation, experimentation, validation, and productionization—at a speed human teams cannot match.
Early signals and enterprise targets
Autoscience points to early milestones, including an autonomous system that produced a peer-reviewed paper at an ICLR 2025 workshop and a silver medal finish in a Kaggle competition against more than 3,300 human teams.
Initial deployments are focused on high-stakes areas such as financial services, manufacturing, and fraud detection, where incremental model improvements can translate into measurable business impact. Yuri Sagalov, Managing Director at General Catalyst, said the firm is backing the company’s effort to scale experimentation and translate research into production systems more efficiently.
Related
