Channel K01 · 2026
Regie
Live demo · built on my own initiative
Every AI action recorded in a ledger, and carried out only after your approval.
Try live (in Dutch)
In short
- Built for
- Businesses that want to use AI employees but keep control
- The problem
- You do not let an AI answer customers on its own without seeing what it does
- What it delivers
- You decide what the AI may do on its own; the rest waits for your approval. Everything goes into a log that cannot be changed unnoticed afterwards. Try it live.
Overview
Regie is a control room for a business's AI employees. The AI proposes replies and actions, and everything it sees, proposes and does goes, in plain language, into a log that cannot be changed unnoticed afterwards. Try it yourself: send a message as a customer, then approve the draft reply as the business owner.
Challenge
AI agents that take action on their own sound powerful, but without a traceable record no business owner dares let them loose on customers. "Just trust the AI" is not a strategy.
What it does
- 01Receive
Every message arrives in one place, whatever the channel
- 02Propose
The AI proposes a reply and says why and how sure it is
- 03Approve
If the system is in doubt, it waits for you: visible, not hidden
- 04Execute
Only after your approval does it really happen, and it is recorded as evidence
Result
- You decide, per type of task, what the AI may do on its own; the rest waits for your approval
- Every proposal and every action is in a log that cannot be changed unnoticed afterwards
- With one button you pause everything and roll actions back
- Try it live: send a message yourself and approve it
Technology
An append-only, hash-chained Action Ledger records every observation, decision and action: tamper-evident and verifiable. A trust ladder sets, per action type, how much autonomy an AI employee has earned; when in doubt, the system waits for human approval. A halt button stops a tenant immediately.