Autonomy you can switch off
— at the identity layer, not just the wallet.
An AI agent must prove it's authorized — with a credential you signed — before it can spend a cent. Overreach is denied. And you can fire it instantly by revoking that credential: it's locked out even though its wallet still holds your funds. Powered by LiberProof. Every spend is a real on-chain transaction.
Meet the agent
It generates its own keypair and DID — that pair is its wallet and its identity. Owned by you, never rented.
You authorize it
You sign a LiberProof credential: "authorized to spend up to 2 tADA, for cloud compute." First we grant the agent an allowance, then you issue the credential.
✦ The agent proves itself
It presents the credential → LiberProof verifies the signature & that it isn't revoked. No valid proof, no payment.
It acts within limit
Gate passes → it autonomously pays the recipient 1 tADA — a real on-chain tx, inside the budget you set.
✦ It overreaches → DENIED
It tries to spend 50 tADA — over the limit. The gate stops it. No transaction.
✦ You revoke it → instant lockout
Pull the credential. The agent tries to act again → its proof no longer verifies → locked out. The wallet still has funds — it just can't act.