platoseedplatoseed
Foundation Models for Scientific Research
We're building the infrastructure for the era of autonomous science. Our core thesis: scientific foundation models with real research "taste" require two things, built in parallel. A product that captures high-quality organic research-process data at scale, and the training and research to turn that data into models with genuine scientific "taste". We're building both.
International Olympiad in Artificial Intelligence 2025 (HM). USACO Platinum division (top 0.01% of competitive programmers globally). Prev: IOAI '25; Z-Fellows; Emergent Ventures
Built COVID-19 helpline infrastructure for Indian govt serving 50k+ people daily Patented (US, SG, IND) and Sold Orthopaedic AISock for 5-figures at 17 Published at NeurIPS, ICML, ICLR, AAAI, CVPR & other A* conferences Prev: aisock (Acq.), Z-Fellows, Emergent Ventures
Cursor for Scientific Research.
Synthetic Sciences offers a platform where AI co-scientists handle the full research loop—from literature review to hypotheses, experiments, results, and publication drafts—for scientific domains, starting with ML and computational biology. It targets researchers and R&D teams seeking to accelerate scientific work by integrating literature, code, and experiments into a unified workflow.
Formerly “InkVell” · why startups rename →
Autonomous AI research
AI models that learn from experience
