Sage
Reads each proposal and explains in plain language what's actually being decided.
Governance where AI helps — and never votes for you.
Agents read proposals, simulate tradeoffs, and execute voted actions. The on-chain policy is strict. The decision is always yours.
Experimental. Testnet only. Not an investment.
Most people in a DAO never vote.
In most DAOs, fewer than 15% of members vote on a given proposal.1 Not because they don't care. Because by the time they've finished reading the proposal — if they finish — the deadline has already passed, or someone else has framed the decision for them.
Governance in DAOs is supposed to be collective. In practice, it's a small group of delegates doing the work, and everyone else trusting them (or not bothering). The structure looks democratic. The outcome often isn't.
The problem isn't that people are lazy. It's that informed participation has been made unreasonably expensive.
Footnote 1.Median participation rates across generalist DAOs typically sit between 5% and 15% per proposal (Snapshot & DAOStack aggregate data, 2024–2025). Sources: coinlaw.io, krayondigital.com. ↩
Two families of agents.
DAOIA is built around two families of agents: those that help members decide, and those that execute what's been decided.
None of them vote. None of them act outside their mandate.
Governance
Help members understand, weigh, and decide on proposals.
Simulator
Tests what would happen if a proposal passes — budget impact, timeline, dependencies — so voters see the consequences before voting.
Advocate
Surfaces the strongest arguments for and against. Flags conflicts of interest. Cites past decisions that set precedent.
Notifier
Alerts members about proposals that match their interests, deadlines approaching, and unusual voting patterns.
Delegate Assistant
Helps delegates stay aligned with what they've publicly committed to — and tells them when they drift.
Execution
Carry out what the DAO has voted on, within strict limits.
Marketing
Drafts external communications. Nothing goes public without a human signing off.
DevOps
Keeps the technical side running — deployments, monitoring, updates the DAO has voted on.
Treasury
Executes financial actions that have been voted on. Reports on fund usage. Refuses to act outside its mandate.
Moderation
Applies community rules across shared spaces. Borderline cases go to humans, not bots.
Research
Analyzes patterns in past votes and proposals. Surfaces insights that inform better future decisions.
How DAOIA works in practice.
A community proposal moves through five steps. Agents do the reading, the math, the watching, and the execution — but never the deciding.
A member submits a proposal
Alice posts: "Allocate 5,000 USDC to sponsor a NYC community meetup in May."
Sage produces a clear summary
Sage reads the full proposal and surfaces a 3-line summary: what's being asked, the budget, the timeline.
Simulator and Advocate add context
Simulator shows what the treasury looks like after the spend. Advocate cites a similar event from Q3 (cost: 7K, attendance: 200) so voters can compare.
Members vote
Each member sees the same summary, the same context, the same trade-offs. They vote based on what they actually understand.
Treasury executes
If the proposal passes, Treasury sends the 5,000 USDC to the recipient multisig and reports back with the transaction hash.
All actions are logged on-chain. Agents act only within policies the DAO has voted on.
Built with limits.
DAOIA is designed so that every agent action stays within boundaries the DAO has explicitly approved. The system enforces this — not just promises it.
- Voted policy — No agent acts outside a policy the DAO has voted on.
- Public audit log — Every action is recorded publicly and inspectable.
- Human override — The DAO can stop, modify, or revoke any policy at any time.
- Responsible disclosure — Bugs and limitations are disclosed within 90 days, patched or not.
The full set of commitments — including what DAOIA will never do — lives in the Principles document.
Join the build.
DAOIA is in Phase 0 — open source, public, and looking for thoughtful early contributors.
Whether you want to test, critique, or just observe — there's a door for you.
Testnet prototype in the works. Not live yet.