Coordinal Research

Coordinal Research is an organization building research automation infrastructure for AI safety and alignment. It was formed through the merger of Lyra Research (led by Jacques Thibodeau) and Vectis AI (led by Ronak Mehta), with the two teams coming together during the Catalyze Impact AI safety incubator program in London. The organization is co-founded by Ronak Mehta, an ML/CS PhD graduate from the University of Wisconsin-Madison and MATS 6.0 alumnus whose research focuses on provably safe AI and formally verified code generation, and Jacques Thibodeau, a data scientist, MATS scholar (2022), and independent alignment researcher who previously created the Alignment Research Literature Dataset used by major AI labs. Coordinal's core technical contribution is an automated research scaffold that accepts any research plan or task and autonomously conducts background research, implements software, evaluates experimental results, and writes research reports. They have also developed Seed, a structured AI-aided research workflow that helps researchers refine ideas and create detailed project specifications. The team has curated over 400 open questions in AI safety for direct use with their system. The organization describes its motivating insight as: the bottleneck in research is not intelligence but taste and verification at scale. Their systems are designed as taste accelerators that distinguish breakthrough discoveries from noise better than traditional peer review, producing research that directly converts into usable products and infrastructure. Coordinal Research, Inc. is incorporated as a for-profit company, with governance structures (such as a Public Benefit Corporation structure) intended to keep it aligned with its mission. They secured approximately $110,000 in seed funding through their Seed Funding Circle and raised additional funds through Manifund, where they signed an MFN SAFE for investor contributions.