Your ERP has the data.
Your team still rekeys it by hand.
We build AI that reads your documents, checks your quality data and supports your planning, wired into the systems you already run. Built by our team in Ghent and Colombo, and put into production rather than left in a demo.
Five things worth automating first
Where the work is repetitive, the data already exists, and the payback is easy to see.
Document and data automation
Orders, invoices and specifications arrive as PDFs and somebody retypes them. We read them, check the fields against what the system already knows, and write them through. Anything the model is unsure about goes to a person rather than being guessed at.
Connects to: ERP, document management, shared mailboxes, supplier portals
Quality and inspection support
The inspection data exists, but nobody has time to look at all of it, so problems surface after the batch has moved on. We rank it, so the right one gets a second look. The output is a list a person acts on, not an automated reject decision taken out of your hands.
Connects to: MES, quality systems, machine and sensor data, historians
Planning and forecasting
Demand and capacity planning usually runs on experience and one person's spreadsheet. We build forecasting on your own order history instead, and show the working so planners can disagree with it.
Connects to: ERP, order history, production planning, procurement
Copilots on your own data
The answer exists somewhere: a spec, a past order, a system only one department can query. An assistant that answers from those sources and cites where it got it, with access scoped by role.
Connects to: SharePoint, Microsoft 365, ERP, internal databases and wikis
ERP and systems integration
The useful work sits between systems that were never meant to talk to each other, some older than the people maintaining them. We build that layer. It is the unglamorous part that makes the rest possible.
Connects to: ERP, MES, WMS, Microsoft 365, bespoke internal systems
Sometimes the answer is not AI
A good share of what gets sold as AI is a database query, a rules engine and a decent form. If that is what your problem needs, we build that instead. It costs less and breaks less often.
We would rather lose the bigger piece of work than put a model in front of something that did not need one. Those get switched off later, and we are the ones explaining why.
From assessment to production
Assessment
We look at the systems, the data and the process, then come back with what is worth doing and what is not.
Prototype
A working prototype against your real data, so you can judge it for yourself.
Production
Hardened, integrated and live in your environment, with monitoring and documentation.
Support
We keep it running and keep improving it. Same team, same contact.
The things buyers actually ask
Does our data leave our systems?
Your decision, not ours. Some work runs entirely inside your environment. Some uses an external model provider, in which case we tell you exactly what is sent and what is retained.
What happens when the AI gets it wrong?
It will sometimes get it wrong, and a vendor who tells you otherwise is selling something. What matters is what happens next: confidence thresholds, a person in the loop where an error is costly, and an audit trail of what the system did and why. Where a mistake is expensive, it escalates rather than proceeds.
Our ERP is old and heavily customised. Is that a problem?
It is normal, and usually the reason the work is worth doing. Most of our integration effort goes into systems that predate current APIs. The assessment tells you whether yours can be integrated cleanly, worked around, or left alone.
Who owns the code?
If you want to own the code we write for you, you do. We agree it at the start of the work, so there is no question about it later.
How long before we see something real?
The prototype stage exists so you are not waiting on a strategy phase to find out whether this works.
Typically about a week from the assessment to a first prototype running on your own data.
What does it cost?
Project based for defined work, a monthly retainer for ongoing capacity, or tied to an agreed outcome. We say which one we think fits before you ask.
If the assessment leads to a project, its cost is credited against that project. If you are not happy with the assessment, you do not pay for it. We do not publish a starting price: it depends too much on your systems and the scope to be a useful number, so you get a figure once we have seen them.
Bring us the process that eats the most hours
We will tell you whether AI is the right answer for it, and what it would take. If it is not, we will tell you that too.
Book a 30-minute call →