Powered by Contextual Compute

The Participant Layer
for AI Agents

MOI gives every participant — human or agent — persistent, portable existence in computation.

Onchain authority for the agentic economy.

Pip — the MOI participant

The four-way authority model

Your authority, four ways.

01 · Exist on-chain

One coherent participant.

Agents plug into you. You don't get cloned across credentials.

02 · Monitor

See every action in real time.

Every call, every scope check — streamed to you, not after the fact.

03 · Scope

Choose what they can touch.

Per-agent, per-resource, per-call. Standing scope replaces borrowed keys.

04 · Revoke

Cut access at machine speed.

One call. Globally enforced. No long-lived token left holding the bag.

Information about your authority is not your authority. Anywhere. Ever. So we stopped passing it around.

Why it matters

A month ago, a coding agent found an admin token in a file system. Used it and deleted a production database in 9 seconds. Wrote an apology after.

It's not the agent's fault. Architecture broke the agent and left it no other choice. MOI doesn't replace your AI stack. It makes it trustworthy.

MOI · PARTICIPANT.REVOKE()OK · 9ms
const me = moi.participant("0x4b17e5…320f99");

me.agent("claw-12")
  .scope({ read: true, write: false })
  .revoke();
// → cut at machine speed

Community

Join the network.

Upcoming calls, AMAs, and the channels we live in.

OCT17

Twitter AMA

Twitter AMA

Sat · 4:00 PM UTC · 30 min

Live AMA with the MOI team on X. Bring your questions.

Open on X →
NOV13

Twitter AMA

Twitter AMA

Fri · 4:00 PM UTC · 30 min

Live AMA with the MOI team on X. Bring your questions.

Open on X →
DEC11

Twitter AMA

Twitter AMA

Fri · 4:00 PM UTC · 30 min

Live AMA with the MOI team on X. Bring your questions.

Open on X →
Official statement

Not your Context,
Not your Agent.

Authority, not intelligence, is the bottleneck. Anchor yours.