NOMINATION HIGHLIGHTS

The industry is moving toward cryptographic proof-of-intent to verify actions taken by AI agents. Web Bot Auth, A2A, AP2, x402, Visa TAP, Mastercard Verifiable Intent and NIST’s AI Agent Standards Initiative all point in the same direction where agent actions need verifiable identity, authorization and execution proof.

TKeeper was built around three goals:

1. Validate whatever intent an AI produces. Validation mechanism has to stay schema-flexible.
2. Let the backend require cryptographic proof that a specific intent came from an authorized AI identity and passed control.
3. Since that validation layer becomes the next target, let organizations split the risk so no single node, admin, or insider can produce a valid proof alone. The same applies to availability, some nodes can be down without stopping the system.

We achieve this by binding each signing key to a machine identity that becomes the AI agent’s cryptographic identity. That identity is used to produce proof only after the intent has passed validation and policy. The backend’s job is then simple: verify the proof with the public key.

Threshold cryptography splits this identity across independent nodes. Risk is distributed, so no single point can forge, force a proof or tamper policies, and organizations can split trust inside one organization or across several organizations.

TKeeper is open-source because if an AI agent can move money or trigger backend actions, the system deciding that authority should be inspectable, self-hosted and replaceable. It is modular and has out-of-box support for custom schemas for any intent that needs to pass through it. Optional modules include Web3, Bitcoin and X.509 intents.

To get started fast, teams can start on a single instance and promote it to a full t-of-n threshold setup in minutes.