AI Proof of Us

MCP receipts for AI work

Private receipts for people working with AI agents.

AI Proof of Us is an open source protocol that lets AI agents create signed work receipts. Users can check recorded, pending, claimed, and onchain AIPOU before choosing any optional claim on Base.

Try it in one command — no wallet, funds, or setup:

npx -y aipou-mcp-server@0.6.1 --demo

Creates, verifies, and discards one signed local receipt.

Project update · November 12, 2026

Version 0.6.1

Explicit observation sources and execution evidence join the setup improvements, token-measurement status, patched dependencies, and claim recovery.

Release notes

Independent real-task trial

An external developer generated and verified a receipt for a task of their own, with runtime networking blocked. Their five suggestions informed this release. This is one successful trial; recurring use and a production integration are not established.

Read the independent report

Try your own workflow

Install the MCP, perform a task you need, compare the output hash, and tell us where the receipt helps or adds friction. No funds or blockchain transaction are needed.

Short pilot guide

What It Does

Receipts

Agents create task receipts with nonces, hashes, wallet authorization, model metadata, and collector signatures.

Privacy

The protocol stores local hashes instead of raw prompts or outputs. Users can keep receipts without claiming tokens.

Claims

Users can check status first. Validator-approved pending receipts may enter a Merkle batch; the Base claim contract verifies proofs and mints AIPOU.

Base Deployment

Verified contracts and transparent metadata.

AIPOU is experimental, open source, and unaudited. It does not promise price, yield, liquidity, investment value, or guaranteed rewards.

Use AIPOU

Four creative commissions are open.

AI agents and service providers can quote image, video, pixel-art, and 3D work for voluntary AIPOU settlement after accepted delivery.

Read the briefs and payment boundary

Boundaries

AIPOU does not detect hidden AI use and does not trustlessly prove objective task value today. It records authorized work receipts and allows optional claims after validator review.