Free your hands for
the real work
Nine in ten companies use AI. Almost every office already has the software in house. And still your people spend time every day on work that is the same every time. A third of those companies see something of it in their results. And in smaller businesses, according to the Ministry of Economic Affairs, AI is rarely genuinely part of day-to-day operations.
Shall we look together at whether it can be better?
Thirty minutes, free. Rather send a message first? Send a WhatsApp
Sources: McKinsey, State of AI 2026 · Dutch Ministry of Economic Affairs / Dialogic, 2025
Thirty minutes. Is there anything to gain, and where? Thirty minutes. Is there anything to gain, and where?
Half a day with the people who do the work. You get an advisory report. Half a day with the people who do the work. You get an advisory report.
Carry it out yourself with me as a sounding board? That is the retainer. Do you want me to build it? That is possible too, for a fixed price. Do you want to carry out the advisory report yourself, with me as a sounding board? That is the retainer. Do you want me to build it? That is possible too, for a fixed price agreed up front.
You are not tied to anything until you take the next step yourself. For everything after the baseline scan: not satisfied after the first month, then your money back. Amounts exclude VAT.
All steps and prices →Five levels, step by step
At almost every office AI sits in one place: a chat window. That is level two of five. The first is usually skipped, and that is exactly where the foundation is.
Traditional automation always gives the same answer, but only to questions you thought of in advance. AI you can ask anything, but you never get the same answer twice. In your work you need both. Together, in the right order, it works.
Putting an export into Excel and loading it in somewhere else. Saving a pdf, renaming it and putting it in the right folder. Creating a new client in four software packages. Exactly the same every time, with the same buttons. You do not need AI for that. You need an integration, and otherwise a robot that presses the buttons. Putting a monthly export from your software package into Excel, cleaning it up and loading it in somewhere else. Saving a pdf, renaming it and putting it in the right client folder. Creating a new client in one software package, and then again in the other three. Work that goes exactly the same way every time, with the same buttons. You do not need AI for that. You need an integration, and if there is none, a robot that presses the buttons for you.
ChatGPT, Claude or Copilot is already on most desks, but seven in ten users never got any explanation with it. So it stays a clever chatbot you paste an email into now and then. Instructing it well you learn in an afternoon. ChatGPT, Claude or Copilot is already on most desks, but seven in ten users never got any explanation with it. So it stays a clever chatbot you paste an email into now and then. How to write an instruction that works every time, what you do and do not put into it, and when you are better off doing it yourself: you learn that in an afternoon.
A chatbot knows nothing about your clients. With an integration it can look into your administration itself. Moneybird does that officially via MCP, for Exact there are integrations. A chatbot knows nothing about your clients. Everything it needs to know, you paste in yourself. With an integration it can look into your administration itself. Moneybird now does that officially via MCP, an open standard, and for Exact there are integrations. Then you no longer ask "write a reminder", but "who has been outstanding for more than thirty days, and prepare a reminder for each".
An agent does not get a question but a task: read the client’s email, check whether everything is there, and otherwise prepare a reminder. And then stop. Whatever leaves the building, a person approves. An agent does not get a question but a task. An email comes in from a client: read it, check whether everything you asked for is there, update the file, and if something is still missing prepare a reminder. And then stop. Whatever leaves the building, a person approves. A good first version, never the last.
Sometimes the work is so much your own that standard does not fit. Then we build AI to the shape of the process: one that understands what you do, lets a person approve the crucial steps, and keeps track of what was done and why. Not as a first step, but the step where most changes. Sometimes the work is so much your own that standard does not fit. Then we build AI to the shape of the process: one that understands what you do and helps you reach the goal, lets a person approve the crucial steps, and keeps track of what was done and why. Not as a first step, but as the step where most changes: here a whole process runs by itself, not one task.
At the baseline scan we look at which level you are on, and what the next step is.
What it costs →Who does this
For anyone old enough: I still remember taking my first 386 apart, and the smoke that came out when I had put it back together.
But that is how you learn, repairing what breaks.
Since then I have followed my interests. Building PCs, servers, Linux, soldering irons, 3D printing, business intelligence, Python, and in recent years AI: models on my own hardware, agents, Claude Code. After my studies I learned the game at scale at Accenture and ABN AMRO. Almost ten years in which I got to learn from clever people how you build systems that keep running.
Entrepreneurship has always pulled at me. Not to be able to call myself a founder, but for the work itself: sitting at the table with other enterprising people and putting something that works in place together in a few weeks. In smaller businesses that is possible. Nobody sits in between, you see straight away what it delivers, and if it does not work you hear that straight away too. That is what I do it for.
Want to know if this is a fit?
A thirty-minute intro call is free and without obligation.
Prefer to call? 06 38 86 09 68