SONATE by YSEEKU

Cryptographic receipts that prove what your AI actually did.

5 followers

Cryptographic receipts that prove what your AI actually did.

5 followers

SONATE creates cryptographically signed Trust Receipts for AI interactions. Each receipt binds the prompt, response, model metadata, policy outcome, and governance context into tamper-evident evidence anyone can verify. Unlike ordinary logs, receipts are portable, independently checkable, and fail verification if altered. Generate a response. Govern it. Break the receipt. Verify it yourself.

Launch tags:GitHub•Vercel Day

Launch Team / Built With

Maker

📌

Hi Product Hunt — I’m Stephen, founder of YSEEKU and creator of SONATE. I built SONATE around a simple problem: AI systems can generate outputs, logs, scores, and dashboards — but those records are usually controlled by the same system that produced them. SONATE creates cryptographically signed Trust Receipts that bind an AI interaction to its governance outcome, so the evidence can be independently verified later. For this launch, I wanted the demo to make that idea tangible rather than just describe it: Generate a response. Govern it. Break the receipt. You can run a prompt through the live demo, inspect the signed receipt, verify it in the browser, then alter a field and watch verification fail. A Trust Receipt does not claim the AI was correct. It proves what was recorded, which policy result was attached, who signed it, and whether the evidence has changed. I’d especially value feedback on three things: Does the concept make sense quickly? Does the demo make the difference from ordinary logs clear? Where would independently verifiable AI evidence be most useful in your work? Thanks for taking a look.

Report

3mo ago

Adding a public dashboard where anyone could browse recent receipts (without exposing sensitive prompts) would make verification way more approachable and build trust faster.

Report

3mo ago

Maker

@ratkca43150

Thanks, Fırat — great suggestion. A privacy-safe public receipt explorer could make the verification model much easier to understand, especially if it exposes only redacted metadata and cryptographic proofs rather than prompts or sensitive content. We’re exploring how to make trust visible without compromising privacy.

Report

3mo ago

Took it for a quick spin and the verification flow actually feels solid — I tweaked a prompt hash and it caught the change instantly. The cryptographic binding of model metadata to the receipt is a clever touch for anyone serious about AI governance.

Report

3mo ago

Maker

@nonayl79703 Thanks for putting the tamper detection through its paces, Necati! Catching exactly that kind of silent modification is what SONATE is designed for. Binding the model metadata and interaction evidence into the signed receipt gives governance teams something independently verifiable—not just another audit log.

Report

3mo ago