When a person retypes data between systems, chases a status or rebuilds a picture from exports at month end, that is cost. When the system does it, that is capacity. AI automation is how you move work from the first column to the second without losing control of it.
Each of these is a delivered build. Clients are anonymised at their request, so the sector, the scale and the measured outcome are real and the name is withheld.
A foodservice wholesaler under £3M turnover
Orders arrived by email, phone and messaging app, 60 to 90 a day, and were retyped into one system by hand.
One queue shows the whole day's work. A person still confirms every order, promises availability and agrees price.
A same day and next day freight operator under £3M turnover
Deliveries finished on the road stayed open in the office while somebody located the proof and reconciled the exception.
Evidence is matched to the consignment as it arrives, so the office is no longer a day behind the road.
A chilled meals producer around £2M turnover
180 to 230 supplier invoices a month, checked against delivery evidence kept in a warehouse folder or on paper.
Invoices are matched to what was actually received. The system never creates a payment or alters a bank detail.
The objective is to remove unnecessary work and increase capacity. It is not headcount reduction, and we would rather say that before you ask than after you have signed.
A model is used where judgement over messy input is genuinely needed, such as reading an unfamiliar invoice layout or reconciling inconsistent supplier feeds. Every figure that follows is computed by deterministic code against a named source. No number in an Orchyn system is written by a model.
Anything with commercial, legal or reputational consequence stays with a person. The system prepares and evidences the decision; a named person in your organisation makes it. Automation removes the work, not the accountability.
The recurring lesson across these builds is that the technology was the smaller half. In the wholesaler, the bottleneck was never typing speed, it was that nobody could see the whole day's work in one place. In the freight operator, fixing reference conventions did more work than the model did.
That is why the work starts with the operation rather than the technology. Automating a process nobody has straightened out first mostly automates the confusion.