How we work
Three weeks, and you can check us at the end of each one.
Week one
You stop explaining the process. We have watched it run.
We sit with the people doing the work: the actual clicks, the actual spreadsheet, the exceptions nobody wrote down. By Friday you have your own process written back to you, and you correct it.
Your part: Two conversations, about an hour each. Nothing to prepare.
Week two
One path works end to end. You run it yourself.
Not a demo and not a slide. The narrowest complete path through the system, live, on your data, in your hands. It will still be missing edge cases. That is why you see it in week two and not at the end.
Your part: Thirty minutes clicking through it, and a list of what is wrong.
Week three
It runs. You have watched it survive a bad input.
Your week two corrections go in. Then we feed it the malformed row, the missing field, the supplier who replies with a photograph, and you watch what it does instead of failing quietly. Logins, keys and ownership move into your accounts in the same week.
Your part: Approve it running on live work.
After
It runs without you asking about it.
Monitoring stays on after handover, so a break is something we tell you about rather than something you discover. Changes later are quoted one at a time.
Your part: Nothing. That was the point.
Ops and procurement builds carry a fourth week. You are told that at the quote, not in week three.
What it is built on
- n8n for orchestration. Supabase for data. Model APIs only where language is genuinely needed, and nowhere else.
- Your existing tools stay where they are: Google Workspace, WhatsApp Business, Meta and Google Ads.
- No proprietary platform. The workflows, the database and the keys sit in your accounts from week three, which means you can fire us and keep the system.
Three weeks from now this is either running or it is not. That is a short enough bet to make on a call.
Start with one call