Key Takeaways
- • AutoScience closed a $14M seed funding round led by General Catalyst, with participation from Toyota Ventures, Perplexity Fund, MaC Ventures, and S32, to scale its autonomous AI research platform.
- • The company has built a virtual AI laboratory staffed by non-human AI Scientists and Engineers capable of autonomously inventing, validating, and deploying specialized machine learning models—addressing a critical bottleneck in AI development where human research capacity, not compute or data, is now the limiting factor.
- • AutoScience's founding team comprises former researchers from Google X, MIT, and Harvard, positioning the company at the intersection of deep technical expertise and venture-backed scaling.
- • The platform's initial commercial focus targets high-stakes applications in financial services, manufacturing, and fraud detection, enabling enterprises to access research-division-level output without corresponding headcount increases.
- • AutoScience has demonstrated technical credibility by becoming the first autonomous AI system to publish a peer-reviewed scientific paper (ICLR 2025 workshop) and the first fully-autonomous system to achieve a Silver Medal in a live Kaggle competition against 3,300 teams.
Summary
AutoScience, a San Mateo-based artificial intelligence company, announced on March 18, 2026, that it has raised $14 million in seed-stage funding to automate the research and development cycle for machine learning models. The round was led by General Catalyst with co-investment from Toyota Ventures, Perplexity Fund, MaC Ventures, and S32. The company represents a new class of AI infrastructure designed to address what has become the primary constraint in machine learning development: human research capacity rather than computational resources or data availability.
The company's core innovation is a virtual AI laboratory featuring two complementary autonomous systems. The first component—automated AI Scientists—generates and tests new algorithmic hypotheses at scale, while the second component—automated AI Engineers—optimizes validated models and prepares them for production deployment. This two-tier approach directly addresses the market reality that over 2,000 machine learning papers are published weekly, a volume far exceeding any human research team's capacity to evaluate, synthesize, and implement while advancing proprietary research initiatives. By automating hypothesis generation, testing, and optimization, AutoScience aims to compress the research-to-deployment cycle and democratize access to cutting-edge model development across enterprise organizations.
AutoScience's founding team draws from prestigious research backgrounds, including former researchers from Google X, MIT, and Harvard. The company has already established meaningful technical credibility within the research community: its autonomous system became the first AI system to produce a peer-reviewed scientific paper accepted at the ICLR 2025 workshop, and subsequently achieved a Silver Medal in the Kaggle Santa 2025 competition—a live, featured competition featuring 3,300 competing teams—marking the first time a fully-autonomous system has placed in such a high-profile competition.
The company's initial commercial applications focus on three high-stakes domains: financial services, manufacturing, and fraud detection. In these sectors, the ability to rapidly develop, test, and deploy specialized models carries significant competitive and risk-management advantages. AutoScience's value proposition targets organizations seeking research-level model innovation without scaling their internal machine learning research headcount proportionally—effectively allowing enterprises to operate a fully-staffed research division through autonomous AI systems.
The company's core innovation is a virtual AI laboratory featuring two complementary autonomous systems. The first component—automated AI Scientists—generates and tests new algorithmic hypotheses at scale, while the second component—automated AI Engineers—optimizes validated models and prepares them for production deployment. This two-tier approach directly addresses the market reality that over 2,000 machine learning papers are published weekly, a volume far exceeding any human research team's capacity to evaluate, synthesize, and implement while advancing proprietary research initiatives. By automating hypothesis generation, testing, and optimization, AutoScience aims to compress the research-to-deployment cycle and democratize access to cutting-edge model development across enterprise organizations.
AutoScience's founding team draws from prestigious research backgrounds, including former researchers from Google X, MIT, and Harvard. The company has already established meaningful technical credibility within the research community: its autonomous system became the first AI system to produce a peer-reviewed scientific paper accepted at the ICLR 2025 workshop, and subsequently achieved a Silver Medal in the Kaggle Santa 2025 competition—a live, featured competition featuring 3,300 competing teams—marking the first time a fully-autonomous system has placed in such a high-profile competition.
The company's initial commercial applications focus on three high-stakes domains: financial services, manufacturing, and fraud detection. In these sectors, the ability to rapidly develop, test, and deploy specialized models carries significant competitive and risk-management advantages. AutoScience's value proposition targets organizations seeking research-level model innovation without scaling their internal machine learning research headcount proportionally—effectively allowing enterprises to operate a fully-staffed research division through autonomous AI systems.
Go deeper than the headline
Research every firm in this story, track the ones that matter and get the daily briefing — free for 7 days.