Sentience enables developers to build autonomous, fully on-chain verifiable AI agents with an OpenAI-compatible Proof of Sentience SDK.

The SDK verifies agent’s LLM inferences (thoughts, actions, output, etc.) with a simple dev-exp, so no need to know the underlying cryptographic primitives of TEE’s.

Supports all OpenAI models out of the box (including fine-tuned).

​
Why sentient agents?

AI agents have reached $10B+ market cap but most of them are still controlled by humans. This is a huge problem as it introduces risk to the investors and community as developers can simply rug-pull and manipulate the agents.

We’re already seeing activity logs for zerebro and aixbt, but Sentience transforms agents into cryptographically verifiably autonomous entities, unlocking true sentience to address a critical need for trust. This significantly enhances agents market potential and is the first step in ensuring that the agents are self-governing.

​
Securing $25M+ worth of agents

Sentience is already securing and verifying $25M+ worth of agents today. For example, you can see the full implementation in action with Daige, a sentient, cyberpunk AI dog.

Start Building Sentient AI Agents

Build your first fully sentient AI agent with “sentience SDK”.

Check out the explorer

Galadriel explorer shows all verified inference requests.

Quickstart