Recently
@a16zcrypto published article about "The missing infrastructure for AI agents: 5 ways blockchains can help" where great contributor were there including
@cc_crowley,
@ccatalini,
@ahall_research,
@liz_harkavy,
@nlevine19, and
@psneville.
And Guess What : we had build this, now looking to make it live soon -
@MOI_Tech - The Participant Layer of Internet, which can solve this and beyond while catering enterprises, web2 apps and web3 dapps.
Authority lives independently at the participant level. so you don’t embed it into every agent, you keep it separate, and you propagate it only when you deliberately choose to. This creates clean, controllable, and revocable control without baking permissions into every AI instance.
Our Moat: Participant-Level State Management
We own the participant layer.
- The source of truth for “who can do what” sits with the participant, not the agent.
- When the participant’s state is set to “participant,” every downstream decision (even by agents) must resolve against that single source.
- Agents themselves remain completely stateless, they carry no persistent authority.
Why This Matters (The Permission Problem Solved) If every agent simply “holds an API key,” you have a nightmare:
- How do you give or revoke permission dynamically?
- How do you enforce it across thousands of interactions without constant re-authorization?
Our model flips it:
- Agents need authority, they don’t own it.
-Authority lives at the interaction level, issued fresh from the participant’s state on every call.
- Revocation is instant and global: change the participant state once → agents lose access immediately. No chasing down API keys.
Result: True Distributed Authority
- Participant = Stateful (owns the rules)
- Agent = Stateless (executes only what it’s handed)
- Authority = Dynamically granted per interaction This gives real safety and trust to AI agents.
Target Use Cases
- Safe Pool of AI agents (secure, auditable multi-agent environments)
- AI Governance frameworks (enterprise-grade control and compliance)
- Enterprise Data Consent & KYC (solve the consent-entity problem cleanly, participants explicitly grant/revoke consent at source, KYC-compliant by design)