The drawings
Every machine starts on paper you can read. We sit with the people who actually do the work and map the process as it really runs, not as the org chart says it runs. Then we agree, in writing: what the machine will do, what it will never do, what it costs, and which numbers must move for it to be called a success. Within the first week you see a working demonstration built on your own business: your services, your prices, your voice.
You know exactly what you're buying before you've bought it.
The build
We build beside your existing systems, never on top of them. The machine is wired into the tools you already use, tested against real scenarios from your actual history, and tuned until it sounds like your business rather than a bot. You watch progress on a live tracker page, the same discipline our build clients get.
Nothing about your business stops or changes while we work.
Live, supervised
The machine goes to work with human approval gates wherever you want them: every action it takes can wait for a yes from your team until trust is earned. We measure daily against the agreed numbers and tune in the open.
Your people stay in charge. The machine earns autonomy; it is never handed it.
The verdict
Thirty days after go-live you get one page: the numbers, against target, in plain English. If the machine earned its keep, we talk about what it earns next. If it did not, you keep the report, the learnings, and your money's worth of honesty.
The pilot stands alone. We earn the roadmap or we don't.
The parts that never change
Fixed price, always
Agreed before we start, in writing. If we scoped it wrong, that's our cost, not yours.
Plain English, always
No jargon in meetings, proposals or reports. If you ever don't understand something we've said, that's our failure and we'll say it again properly.
Your keys, your machine
Everything we build runs in accounts you own. Leave whenever you like and it keeps working. We think you'll stay because it's good, not because it's locked.
Approval gates by default
The machine asks before it acts, for as long as you want it to. Autonomy is granted by you, gradually, with evidence.
Success has a number
Every engagement names its numbers before the build starts. Vague wins don't count.
We sign our work
The old smiths struck a touchmark into everything they forged, a signature of accountability. Ours is on every machine we build.
The questions everyone has
How much does it cost?
A fixed price, agreed in writing before we start, plus a monthly run fee that covers monitoring, tuning and support. No hourly billing, no open-ended retainers, no surprises. The exact number depends on the machine, which is why the first conversation is free and specific.
How long until something is actually running?
The first machine is typically live in two to four weeks. You see a working demonstration on your own business data inside the first week, before you have committed to anything past the pilot.
Will this break the systems we already use?
No. We build beside what you have, never on top of it. Your existing tools keep working exactly as they do today; the machine connects to them and does the chasing, answering and moving that nobody had time for.
What if it doesn't work for us?
Every engagement has success numbers agreed before we build, and a 30-day verdict after go-live: a one-page report of numbers against target. If the machine did not earn its keep, you keep the report and owe nothing further. The pilot stands alone.
Do our people need to be technical?
Not at all. Everything is explained in your language, every action the machine takes can pass through human approval until you decide otherwise, and you can call and ask anything at any time.
Will AI replace our staff?
The machine takes the work nobody should be doing at 9pm: missed calls, paperwork, chasing, triage. Your people keep the work worth doing. Every client so far has used the recovered hours for more work, not fewer people.