Process
How We Work
Five steps, in this order. Most of what goes wrong on AI projects is caused by doing step four before step one.
The build is the shortest part. It is also the only part anyone imagines when they picture this work.
We analyse the day-to-day operations
Not the part you flagged. The whole company, because owners are usually right about the symptom and often wrong about the cause.
This is where the uncomfortable finding shows up. The operation you were going to automate is third on the list, and the one at the top is held together by two long-serving people and a spreadsheet.
What you end up with: A written list: every operation, whether it can be automated, and what it is worth if it is.
We pick the one that costs the most
Not the whole operation. A slice that can be live in weeks and shows up in a number you already track.
In quoting, that is usually the standard configurations rather than the complex ones. The complex ones stay with a person, on purpose.
What you end up with: A scoped first build with a number attached to it.
We train it on your business
Your quotes, your replies, your job records. Not a generic prompt and not your help centre, because the gap between published policy and actual practice is exactly where these systems fail.
Resolving that gap is a series of decisions only you can make. It tends to improve the manual process on its own.
What you end up with: A system trained on your material, with escalation rules written down.
We build it, or you build it from the plan
Everything lives under your billing from the first week. Where existing tools work, they stay, because rewriting a working connection spends your money to arrive at the same place.
Then it is tested against the last few hundred real cases, where the correct answer is already known because your team produced it.
What you end up with: A live system, with a measured agreement rate against real historical cases.
We train your people to run it
This is where projects land or quietly do not. A system with no owner has a shelf life of about a year regardless of how good it was on day one.
There is a specific week, usually two or three after go-live, when the team stops double-checking every output. Before that week the system is a cost. After it, the number moves.
What you end up with: A trained owner, a written runbook, and no dependency on us.
Read next
Find out what your most expensive operation is
The assessment covers the whole company: how you get customers, how you deliver, and how the place runs. You get it in writing, and you can build from it with or without us.