How we use AI

Automation with a human in the driver’s seat.

Where IDS uses automation and where it doesn’t — plainly, so a regulated-goods operator can trust it.

Book a walkthrough Security & compliance

Our principle

Automate the boring work. Keep judgment with a person.

We automate the boring, no-judgment work. Anything that requires judgment is suggest-and-confirm — the software proposes, a person decides. A human stays accountable for every custody and compliance decision.

Drawing the line

Where we use it — and where we don’t.

Where we use it

Importing and mapping messy CSV/EDI files
Drafting reports for a human to verify
Helping you search your data in plain language
Diagnosing a reported fault against the code and the device record
Writing the fix, for one of our people to review before it ships
Speeding up support so answers come faster

Where we don’t

No autonomous decisions on custody, compliance or who gets a delivery
No black-box automation of a regulated handoff
No fix reaches your system without one of our people reviewing and releasing it

Support & fixes

Diagnosed in minutes. Fixed under supervision.

When something goes wrong in your operation, the clock that matters is how long you wait for a fix. We use AI to collapse the diagnosis, and we keep a person on the release.

The evidence comes with the report — the driver app keeps a sequence-numbered record of what happened on the device — including whether it had signal — and a driver sends it to support in one tap and reads back a short reference code. We start from what actually happened, not from a description of it.
AI reads the code and the record together — instead of a developer spending a day reproducing a fault, automation searches the source and the captured evidence and comes back with the specific cause — the line, not the theory.
A person reviews every change — nothing reaches your system because a machine thought it was right. A fix is written, reviewed by one of our people, and only then released. Nothing is committed or shipped without a human approving it.
Every fix becomes a permanent test — a defect we have seen once is written into the test suite so it cannot come back quietly. Thousands of these checks run against every change — the mistakes of the last twenty years are now guardrails.
The whole suite runs before anything ships — a change that breaks something elsewhere does not get released. That is how a small team ships fast without shipping damage.

What it means for you

Faster fixes, fewer regressions.

An independent vendor is supposed to be the one that cannot keep up with a well-funded competitor’s support desk. Automation is how we invert that — and the human review is why speed does not cost you stability.

It is the same principle as everywhere else on this page: the machine does the boring work, a person makes the call.

Data handling

How we handle your data.

We minimise what’s sent to any model — only what a task genuinely needs.
We don’t use your operational data to train third-party models.
Heavy processing can run on your own key — so it stays under your control.

This page describes our current approach and is being finalized with legal review — ask us for specifics on your data.