Day one of Logproof

Today I started building Logproof. A run log, an approval tap and a client receipt for AI automations that non engineers run.

The problem I keep seeing: a small business in the US, Australia or the UK pays someone, often offshore, to build an n8n, Make or Zapier automation with an AI step in it. It works. Then the owner has no record of what it did, no way to approve a risky action before it goes out, and nothing to show a client or a boss when they ask what the AI touched. The builder moves on. The owner is left running a system they cannot see.

Logproof is one HTTP call per run. The owner gets a log, a pending approval for anything marked risky, and a read only receipt page they can send to anyone.

The playbook is a portfolio of small bets with kill gates. This one gets seven days to have 20 real conversations, no favours from friends, and 30 days to reach three paying customers or 500 dollars. If it misses, it dies and the next bet starts. I will post the numbers either way.

Day one numbers: 0 users, 0 conversations, 0 dollars, one working product, live since tonight.

If you run automations someone else built for you, or you build them for clients, I want 20 minutes with you. The contact details are on this site.