Week 1 update: 0 signups, a working demo, and what I learned about the gap between vision and reality

TrustLoop — Week 1 update: 0 signups, a working demo, and what I learned about the gap between vision and reality

Posted my first IH post last week. Zero signups since. Here's what I've been doing instead of refreshing the waitlist page.

What I shipped this week

I recorded a live demo. Not a slideshow. Not a Loom walkthrough. The actual proxy running, intercepting a real tool call, blocking it, logging it.

You can watch it here: TrustLoop

The 30-second version: an AI agent calls delete_all_files. TrustLoop intercepts it before it executes. The dashboard shows the block in real time. That's the whole pitch in one clip.

What's actually built vs what I'm selling

This is the part I want to be honest about, because IH is the right place for it.

The proxy works. The interception works. The kill-switch works. Plug it into Claude Desktop via MCP and it genuinely blocks calls and logs them to SQLite in real time.

The blockchain anchoring? It exists. It SHA-256 hashes the audit log and writes it to a smart contract. But right now it runs on a local Hardhat test node, not a real chain. And you have to trigger it manually.

So the honest state is: the governance layer is real. The immutability story is a demo skeleton.

What needs to happen next: per-event hashing in the proxy, auto-anchoring to a public chain (probably Polygon or Base), self-serve onboarding so someone can get an API key without talking to me.

The thing nobody tells you about building AI tooling

The hardest part isn't the code. It's that the people who most need this product are the same people who are moving too fast to stop and think about governance.

Every founder I talk to says "yeah we should probably have something like this." Then they go back to shipping features.

I think the only way to sell this is to make the demo so viscerally obvious that they feel the problem in the room. Hence the video.

What I'm focused on this week

• Get one real user. Not a friend. A stranger.

• Wire up Stripe so early access has a price attached

• Move anchoring from localhost to a real testnet

If you watched the demo — what would make you actually pay for this today? What's missing?

Soji Mathew Joseph