Case study
Gov AI
An AI assistant for DAO governance proposals and value-based decision filtering.
Review a proposal before voting
Gov AI takes a DAO proposal URL and produces a structured review: summary, changes, benefits, risks, unknowns, and a recommendation based on the user’s principles. The README lists Snapshot, Tally, DAO DAO, and Mintscan.
Missing information should be marked UNKNOWN. The system is not supposed to invent voting options or results. Its recommendation supports a person’s decision rather than replacing it.
Interfaces and verification
The project supports CLI and API use, with a Telegram bot and report viewer layered over the analysis. The public website exposes saved reports.
The repository also documents model, routing, and repeatability experiments. A deterministic fixture test does not prove model truthfulness, and provider metadata does not automatically verify every research conclusion.
The case connects web3 research with configuring an agent for a bounded process: reviewing a proposal, recording unknowns, and preparing a reasoned decision.