We embed with the person doing the work, build the AI employee with them, and prove it’s right against a written acceptance test. $7,500 fixed per workflow — half up front, half only when the test passes in live use.
The best automation opportunities are invisible from outside — so the pod starts by sitting where the work happens.
We sit with the person actually doing the work — the WhatsApp threads, the booking system, the spreadsheet — and write down how the workflow really runs, not how the org chart says it runs.
Together we pick the exact slice to automate first: high repetition, real dollars, data we can reach. And we write the acceptance test — the pass/fail bar the operator must clear before you pay the second half.
The operator gets built next to the person who owns the workflow, in their words, on your systems. They shape it; they’re not handed it.
It runs on real work, supervised, with every outbound message held for approval. We check it holds up across the messy cases — not just the demo-friendly ones.
The acceptance test runs against the live trial. It passes, you pay the second half, and the workflow is off your plate. It doesn’t pass, you don’t.
The software underneath is open source and cheap — we say so out loud. What you’re paying for is your business encoded into an operator you own, the proof that it’s right, and someone who answers for it.
WhatsApp, Gmail, Google Calendar, your booking or property-management system, your CRM, Stripe, Notion, spreadsheets — and the odd custom system every real business has. If it has a way in, the operator can work it.
Two weeks, one workflow in production. 50% to start, 50% only when the written acceptance test passes in the live trial. Your $1,500 audit is credited against it.
Cheaper than the first, because your business is already written down and the proof harness already exists. Most of the setup work never has to happen twice.
What the machine costs (not much). The operator runs on a Mac mini you buy and own (~$600–1,500 one-time) or a small cloud server (~$30/month) — passed through at cost, because it’s cheap plumbing, not the product. Expect roughly $30–150/month in AI model usage for a working workflow, with hard caps and alerts so it can never run away. Ongoing accountability is its own offering — see the operations retainer.
The $1,500 read-only audit finds the workflow worth automating and the dollars it leaks — and it’s credited in full toward the pod. A read-only look can’t make things worse.