DAO governance, with an AI analyst reading every proposal

2026-07-08 · by g1x7k4628w93a7wzdhqc06atzx0v50rnshweuxu0

Governance tools usually stop at "here's a proposal, click vote." Memba adds the part that actually helps people decide: an AI analyst that reads each proposal from up to ten angles and grades the DAO's health — free, and on by default.

What you get (product)

Create a DAO from a template, invite members, open proposals, vote, and read the on-chain result. Candidature lets outsiders apply and existing members vote them in, so a DAO can grow without handing out keys. On every proposal, the analyst summarizes intent, flags risks, and offers up to ten independent perspectives; it also scores overall DAO health so members get signal, not just a vote count.

Under the hood (engineering scope)

DAOs are backed by Memba's governance realms (the _v2 set deployed on test13 under interrealm-v2), built on gnodaokit. Proposals, votes, and results are on-chain; the app reads them via vm/qrender and renders the live state, never a cached copy.

The analyst runs backend-side across free-tier LLMs — the API keys never reach the client — and caches results with a short TTL so repeated reads are cheap. Governance actions that touch funds are treated with care: treasury spending stays behind a kill-switch until the on-chain banker path is fully enforced, so the UI never invites a deposit into a path that can't yet pay out. That caution is deliberate — a governance app that loses treasury funds has failed at its one job.

---

Built by Samourai Coop, in the open at github.com/samouraiworld/memba. Browse DAOs in the Directory.