Fulcrum blog · Jun 2026
Research post introducing inverse rubric optimization (IRO) as a testbed for agent science, with open-sourced code and findings on long-horizon agent behavior.
Company research
Public company, workplace, funding, and market signals
Updated Jul 29, 2026
Overview
Fulcrum Research is a San Francisco-based AI research lab / developer-tools startup building an agentic debugger for AI systems. Public materials describe it as helping teams debug, red-team, and evaluate AI agents and RL environments; it also publishes open-source tooling such as Lunette, Druids, Quibbler, and Orchestra. The public company site currently redirects, but YC, GitHub, LinkedIn, and the blog consistently tie the company to Fulcrum Research. No separate public CEO title was found; the co-founders are the visible leaders.
Primary product
Agentic debugger for AI systems for monitoring, red-teaming, and debugging AI agents and RL environments
Founded
2025
Headquarters
San Francisco, California, United States
Team size
1-10
Work style
Hybrid
Industry
Technology, Information and Internet