How we work
From the first conversation to year two.
You get a straight answer, and it costs you nothing.
Forty-five minutes with the two people who would actually build it.
01 / 06 · a straight answer
You get a straight answer, and it costs you nothing.
Forty-five minutes with the two people who would actually build it. No deck, no discovery phase you pay for, and nothing to sign at the end of it.
- You are talking to the two people who would write the code. There is no account manager, no qualification call, and nobody hands you over afterwards.
- We ask who outside your company reads your data and what you have already promised them. That answer shapes the architecture more than any feature list does, and it is far cheaper to know now than to discover in year two.
- Sometimes the honest answer is that you need one change and not a partner, or a tool you can buy for a fraction of the cost. We would rather say that in the first hour than in month three.
02 / 06 · every option
You see every option, including the one we would talk you out of.
Two or three working sessions before anything is billable. The real directions side by side — keep it and repair the worst of it, rebuild the core, replace the lot — each with what it costs and what it risks.
- The ceiling you keep hitting is usually one component, not the whole system. Working out which one is most of the value, and it happens before there is a contract.
- We check against your live systems before we quote, not after. On one integration that meant pulling real orders and mapping every status combination first, so the scope carried no guesswork into the build.
- Looking properly often turns up something you already own. On one build it was tens of millions of historical matching decisions nobody had used, which became the training data for the system that took the manual work over.
03 / 06 · what you sign
You sign something you can stop.
Phased, with the effort against each line and at least one thing we are explicitly not doing. What you commit to is a working rhythm you can judge every week, not a scope written before anyone had opened the system.
- Each phase carries a go/no-go point that is yours to make. On one engagement the first phase ran entirely at our risk, with no commitment required to start it.
- A scope agreed before anyone has opened the system is a fiction both sides then spend the engagement defending. You judge us weekly instead — the same slot in your calendar, thirty minutes, a working meeting rather than a status report read aloud.
- It is written down in both directions. What we commit to, and the four things we need from you: one decision-maker who attends, reviews inside a few working days, access to the systems we have to integrate with, and honesty about the budget.
04 / 06 · you own it
You own all of it from day one, not at the end.
Your accounts, your repository, your infrastructure, your intellectual property — created in your name in week one rather than built in ours and migrated to you later. If we disappeared in week two, you would still have all of it.
- That is not a promise about our character; it is a property of whose account the thing runs in. One platform we built runs in production today on the client's own hosting and database projects, not ours.
- The processing agreement is signed before any real data moves. On one integration the whole legal package, NDA and sub-processor agreement included, was signed before a single API call touched production data.
- No hostage situation at renewal and no licensing surprise. Digitale Kliniek was rebuilt from the ground up with the intellectual property staying 100% with the company that owns the product, in its own source control from day one.
05 / 06 · you decide what ships
Nothing reaches your users unless you put it there.
Every change waits in your own environment until you have walked it and approved it. That takes minutes, from a link, whenever suits you — and production only ever contains what you pushed yourself.
- It arrives explained: what changed, how to test it, why we did it, and an honest read on the risk — pointed at on the interface itself rather than buried in a changelog you have to map onto your own screen. Nothing is offered to you at all until it has passed type checks, tests, a migration rehearsal against real production shape, and a tenant-isolation check.
- Raising a problem costs you seconds. Point at the thing on your screen, write a line, and it lands with us exactly as you saw it — you never write a ticket, reproduce a bug, or explain which browser you were in.
- Your call wins, and it is reversible. When we disagree we say so once, with reasoning, and then build what you decided; if something does break in production we roll back first and explain afterwards, because that rollback was rehearsed before the change was ever offered to you.
06 / 06 · year two
In year two, you still have the same two people.
Going live is the middle of the engagement, not the end of it. Every system we have shipped is still running and we still run it — monitoring, incidents, dependency and security updates, and the roadmap conversation that keeps it from going stale.
- Four partners a year is a scheduling fact, not a boast. The weekly session and the operating commitment are only physically possible at that number, which is why the people in your first conversation are the people still answering in year two — no subcontracting, no offshore team you meet after signing.
- When somebody outside your company starts checking, the answer already exists. Who raised it, who approved it, who pushed it and when is written as a by-product of working, in an append-only trail that cannot be edited afterwards by an administrator, by you, or by us.
- That trail is built for a real assessment, not a tidy-up beforehand. On Digitale Kliniek every document, approval, incident and access event is logged to the standard a NEN 7510 audit reads; on an AI build every model call is logged, so an output produced last month is still reproducible after a model upgrade.
