The questions we hear most

Yes, provided there is enough repetition. It is not the number of staff that decides whether automation pays off but how much recurring work passes through every day. Five people who together retype forty requests a day have a better business case than fifty people with no repetition.

Rule of thumb: from around ten hours of manual work a month on a single process it becomes worth working out.

A small automation starts at €995: one clearly bounded step between two places — a form that lands validated in your CRM with a confirmation email going out, for instance. Maintenance starts at €95 a month.

If you want to know first whether it pays off, an Automation Check costs €295. If you have us build within thirty days, we deduct that amount in full.

No, and that is precisely what we do not do. Your ERP, CRM, accounting and planning stay exactly where they are. We build the links between them, because that is where the manual work sits — packages are strong within their own domain and weak at the edges.

A migration or a package replacement is a different trade. If you ask for that, we will point you elsewhere.

No. Everything that can be done with fixed rules is done with fixed rules: they are predictable, testable and cheaper. Pulling an amount out of a fixed invoice layout, matching fields, passing something on — no language model is needed for that.

We use AI where interpretation is needed: reading a document laid out differently every time, classifying free text, or summarising. For critical actions there is always a check on top.

The design is built around that. Every extracted field gets a confidence score; if it stays below the configured threshold, processing stops and a person decides. That is not an emergency brake but the normal behaviour.

On top of that, every processing run is visible in the log: which model, which input, which outcome. If something goes wrong it can be reconstructed rather than guessed at.

Yes. Maintenance is not compulsory here. If you choose to do it yourself you get the workflow, the source code, the documentation and the instructions for deployment and recovery. We charge half a day once for the handover.

We usually do recommend Managed maintenance, but that is experience rather than a condition: it is the third-party integrations that break, not our code.

It happens, and it is the most common reason an automation stops — an integration where the supplier changes something, for instance. Monitoring usually notices before you do: if throughput drops or errors come in, we get an alert.

Under a maintenance contract the repair is covered. If you maintain it yourself it is down to your own people — which is why we hand over integrations documented.

By default in your own environment, on your own licences and accounts. If you would rather not arrange hosting yourself, it can run on our infrastructure; even then the environment stays transferable.

Where a process needs a language model, the default is that we run our own, inside Europe. If an external service is needed, the proposal says which one, what for and where the processing takes place — before we build. No data goes to a non-European service unless that is written down. Your data is not used to train models.

You do. What we build for you is yours, including the source code and the documentation. There is no licence construction around it tying you to us.

That is a deliberate choice in the design and not a reassurance in a meeting: everything runs on your own accounts, access credentials are kept separate, and documentation is written as we build.

From €95 a month for Monitor — monitoring, logging, updates and repair when something in our workflow fails — up to €695 and above for Business, with several connected workflows and higher volumes. In between sit Care (€195) and Managed (€395).

Which tier fits depends on the number of processes and the volume, not on how many integrations we happened to build.

The infrastructure we manage is in the price. If your process uses an external service — a language model, an OCR service, a paid API — we discuss those costs in advance and estimate monthly usage.

Where possible you work with your own accounts and keys: you see exactly what you use, pay the supplier directly, and there is no margin from us in between. We never promise unlimited use; where it can run up, we agree a ceiling with an alert before that ceiling.

A first conversation is usually possible within a week, an Automation Check within two. After that a small automation is live in one to two weeks and a standard process in two to four, counted from the moment we have access to the systems and test data.

In practice that access is more often the limiting factor than the building.

Then we say so, and build nothing. Automation has a fixed cost side: building costs money and maintenance costs money every month. Below a certain volume that maintenance costs more than the manual work it replaces.

That is not an edge case but a common outcome. It is exactly why the analysis is paid: it means we can afford to say no.

That depends on whether the package offers an integration, an export or an import. Most modern packages do — Exact Online, AFAS, Teamleader, HubSpot, Microsoft 365 and Google Workspace, for example. With older systems an intermediate solution is sometimes needed.

This is one of the first things we establish, because it determines what is feasible. Sometimes work is needed at your own software supplier to open up an integration; those costs fall outside our fixed price.

Often it can do something. The reality is that packages are strong within their own domain and weak at the edges between them — exactly where the manual work sits. An ERP can process an order, but not read a customer's PDF, translate part numbers and route an exception to the right colleague.

In the Automation Check we first look at whether your own package can already do it. If it can, we say so, and that €295 is all you pay us.

That is why everything runs by default in your own environment and on your own licences, with documentation written as we build. If the relationship falls away, everything keeps running and another party can take it over without making you start again.

That is a deliberate choice in the design, not a reassurance in a meeting.

That is why we build the error handling before the features. When in doubt a case goes to a person instead of through, and every processing run is visible in the logging.

For the first weeks a new process also runs with an extra check alongside, which you wind down yourself once you trust it.

As little as possible. The starting point is that the work disappears, not that a screen is added. Usually everyone keeps working in the same systems and the only difference is that part of the data entry has already been done.

Where something is added — an exceptions list, for instance — we explain it to the people who work with it, not only to management.

Sometimes that is the right answer, and then we say so. But administrative staff are hard to find, and the Dutch employment agency UWV notes that these roles in particular are being automated further. An automation that is running also works in the evenings and during holidays.

The maintenance tier depends on the number of processes and the volume. If that grows, you move to a higher tier — but we recalibrate the amount in advance, so you are never faced with an increase after the fact.

Yes. A first conversation and an Automation Check work fine remotely, and so does most of the building — the systems are reachable over an API.

For an Automation Scan we prefer to visit, because we need to speak to the people who do the work. Within about an hour's drive of Zoetermeer that is straightforward; further afield we agree in advance how to arrange it practically.

Then we stop and discuss it before building on. A changed scope can mean it costs more, but you hear that in advance and not on the final invoice.