Maingen
Infraday-curatedSan Francisco, California, United States
Simulation environments and benchmarks for training and evaluating AI agents on industrial operations.
Overview
Maingen develops simulation environments and benchmarks for training and evaluating AI agents on industrial operations. The San Francisco company works with operators to translate decision processes and tacit operational knowledge into reinforcement-learning environments covering power plants, factories and data centers. Its first public benchmark, SolarBench, tests whether agents can manage a solar portfolio, and the company says it is expanding into additional industrial settings. Founded in 2026 by Phillip Yan and David Yang, Maingen joined Y Combinator’s Summer 2026 batch and lists a two-person team. Public materials target frontier AI laboratories and industrial operators but do not identify customers or disclose annual revenue.
Infrastructure sectors
Capabilities & services
Operator-workflow modeling
Long-horizon agent evaluation
Industrial decision-task design
Solar portfolio benchmarking
Startup solution
- What they supply
- Software & AI
- Relevant buyers
- Industrial operators and AI research teams
Profile information
Infraday-curated. Claim status confirms profile management rights and does not imply independent verification.
Last editorial check: 2026-10-11.
Represent this organization?
Sign in to submit a claim using your work email. Infraday checks your affiliation before approving profile management.
Sign in to claimSuggest a correction or request removal
Tell us what needs correcting and include an official source. Your email is private.