Trust Model

Don't trust. Verify.

We know that Web3 users are inherently skeptical of software that handles sensitive data, AI prompts, or cryptographic keys. Privacy claims are worthless if you have to take them on faith. NOEMA's trust model rests on architecture first, and on open code exactly where openness is what makes the product work.

Verify by Architecture

The Noema Client runs in your browser, and that has a consequence most privacy products cannot offer: you can watch it work. Open your browser's developer tools during a local session and look at the network tab. When you run a local model, your prompt generates no outbound requests. Local means local, and your own browser proves it, live, on every message. No audit report required, no trust in us required.

What Will Be Open Source

We open components where openness is the point, not as a slogan:

• Provider Node. Operators stake real value on hardware running our software. Nobody should stake into a black box, so the node CLI will be fully open-source and auditable before a single token is staked.
• Noema SDK. A developer tool you can't read is a developer tool you can't trust. The SDK ships open on npm.

The Protected Core

The Noema Client and its routing Engine stay closed while the product matures. Their privacy guarantee doesn't depend on you reading the code; it depends on where the code runs, which is your device, verifiably, as shown above. Openness is reserved for the places where it is what earns trust: the Provider Node you stake into, and the SDK you build on.