Autonomous AI R&D system that proposes, runs and iterates on experiments in parallel
intology.ai/blog/automating-ai-rd-for-molecular-discovery
Locus is Intology's autonomous discovery system for AI R&D problems. It formulates experiment ideas, executes several experiments concurrently on purpose-built infrastructure, and samples new experiments based on evaluation feedback. Intology has applied it to language model post-training, ML research engineering and molecular discovery (property prediction, Boltz-2 potency fine-tuning, antibody design workflows), reporting higher performance than expert baselines and Codex/Claude Code under matched compute (company-reported).