Edison Scientific raises $70M to scale Kosmos, its autonomous AI scientist

Edison Scientific raised $70M to scale Kosmos, an autonomous AI scientist for literature, data analysis, and cited reports. It speeds research in drug discovery and materials.

Edison Scientific raises $70M to build autonomous AI scientists for research

Updated: 18:15 EST - December 18, 2025

Edison Scientific, a San Francisco AI startup spun out of nonprofit Future House Inc., has secured $70 million to scale an autonomous "AI Scientist" platform built to accelerate research. The company plans to expand operations and product development, turning internal research into a broadly usable system for labs and R&D teams.

What the company is building

At the center of Edison's approach is Kosmos, an AI Scientist that can read thousands of papers, run analyses, synthesize findings and generate fully cited reports in one automated run. It coordinates specialized AI agents in parallel and maintains a structured model of the problem space, allowing it to reason across literature and large datasets.

The platform aims to automate large segments of the research workflow end-to-end. That includes literature synthesis, data analysis, molecular and materials design, hypothesis generation and experimental planning-inside a single environment.

How Kosmos fits into real research

Literature to insight: Scans and summarizes massive corpora with citations, cutting weeks of reading into hours.
Hypotheses plus plans: Proposes testable ideas and drafts computational experiments with clear next steps for wet-lab validation.
Data-driven iteration: Runs analyses, updates its model with new evidence and refines recommendations.

This is especially useful in drug discovery, genetics and materials science where cycle time matters. Faster iteration over hypotheses and empirical results can translate directly into pipeline velocity and cost savings.

Why this matters for labs and R&D leaders

Speed: Compresses months of expert effort into days without sacrificing citation trails.
Coverage: Reduces missed signals by scanning across disciplines and heterogeneous datasets.
Repeatability: Codifies steps and decisions, improving handoffs across teams and vendors.
Focus: Frees human experts to make judgment calls while the system handles the grind.

The founder view

"Science is too slow," said co-founders Sam Rodriquez and Andrew White. "At Edison, we are integrating AI Scientists into the full stack of research, from basic discovery to clinical trials. We want cures for all diseases by mid-century."

Funding details

The $70 million seed round was led by Triatomic Capital LP, Spark Capital LP and a major unnamed U.S. institutional biotech investor. Additional participants include Pillar VC LP, Susa Ventures Partners LP, Striker Venture Partners LP, Hawktail Management, Olive Tree Capital Advisors and several angel investors.

What to watch next

Validation: Peer-reviewed benchmarks, prospective studies and real-world case reports that show measurable gains in time-to-result and hit quality.
Integrations: Deeper links to ELNs, LIMS, high-throughput screening and robotic labs to close the loop from planning to execution.
Compliance and IP: Auditability, data governance and secure handling of proprietary datasets for pharma and materials customers.
Partnerships: Deals with CROs, biotechs and industrial R&D groups that put Kosmos into production pipelines.

Bottom line for scientists, operators and investors

Scientific work is becoming a software problem: structured knowledge, iterative computation and fast feedback. Edison Scientific is betting that an autonomous system coordinating many specialized agents can shrink cycle times without losing scientific rigor.

If they execute, expect shorter paths from question to validated result-and more teams running serious research with smaller headcounts and clearer audit trails.