Build
One process that genuinely runs.
No prototype, no pilot that never reaches production: a flow connected to your systems, with a defined scope of action, doing its work every day from sign-off onwards.
Why we go one at a time
A mid-sized company that starts with a large AI programme has a presentation after six months and no running process. The reason is rarely the technology; it is simultaneity. Ten processes to record, ten sets of access to sort out, ten sign-offs to coordinate - and nobody has time for that alongside the day job.
So: one after another. The first costs the most, because it is the one that establishes access, data groundwork and infrastructure. From the second onwards that is already in place, which is why it is cheaper - not a volume discount, but the real cost structure.
You can stop after any process without what came before losing its value. That is the whole point of this order.
What a process project contains
Integration
The systems involved get connected - ERP, inventory, mailbox, file storage. Where no interface exists, we build one. That is the most demanding part and the one everything else fails on.
Scope of action
What the system does itself and what a person signs off is fixed in writing in the specification. Anything binding or irreversible stays subject to approval.
Acceptance against test cases
Checkable criteria agreed BEFORE the build, measured on real past cases from your own operation. Not "works well", but a figure you can accept against.
Logging
Every output stays traceable to its source. Without that you cannot settle, in a disputed case, why the system decided the way it did.
Introduction
The people who will work with it get shown how - on their own cases, not on an example.
All of that is included in the price and never charged separately. It is the part without which a system should not go into production.
Frequently asked
How long does a process project take?
Depending on scope, two to eight weeks to the first production run. The build itself is usually the shorter half; the time goes into access, coordination and test cases.
Can we change the scope mid-project?
Yes, as long as the acceptance criteria are not yet fixed. After that a change becomes a small separate order - not on principle, but because otherwise nobody can say what is being accepted against.
What happens after sign-off?
The process runs. Whether we operate it or you do is your call - though without somebody watching, an AI flow quietly gets worse.
Let us start with a conversation.
You tell us where your time goes. We put it in writing: what is worth automating, and roughly what that costs.
