Cut through the noise
I separate the use cases that will move your business from the expensive distractions. That includes a confident no on the ideas that won't.
AI strategy, training, and execution for ambitious organisations in the Netherlands
Most AI ideas are not worth building. I find the ones that are, then help your team build them.
You get a trusted advisor and a hands-on AI engineer in the same person. Management gets a plan they understand, your developers get someone who has shipped AI agents, RAG pipelines, and production data science himself.
The sentence I hear most
“AI is changing incredibly fast. I want my company to keep up and stay ahead of the competition, but I lack the expertise to steer in the right direction.”
That expertise can be borrowed. For a few weeks or a few months I pick the ideas worth building with you, get the first one built with your own people, and leave them able to build the next one without me.
I've built what I advise on
My name is Guus Bobeldijk. Years of doing exactly this work: as a data-analytics consultant at PwC across banking, telecom, and public organisations, and hands-on as a data scientist and AI engineer. I built and launched AI products solo (RAG, GraphRAG, the lot), so I know first-hand what ships and what only demos well. And because I teach at Nyenrode Business University and train teams in AI for a living, I can sit with your engineers and your director in the same hour and make each understand the other.
- 8 years in data & AI
- Former PwC data analytics consultant
- Lecturer at Nyenrode Business University
- Shipped AI products myself
Trusted by
What I do
Make it buildable
Every idea is scored on impact, effort, and whether your data can actually carry it.
Upskill your team
If you have your own developers, I train them to build the top use cases themselves. The training starts with AI-assisted coding.
AI TrainingBuild it with you
As fractional AI lead I design the system, build the first version with your team, and hand it over. The same goes for less technical work, like getting your people to actually use the Microsoft 365 Copilot licences you already pay for.
Four ways in
Every engagement stands on its own and delivers value by itself. Each one also opens a natural next step.
AI Quickscan
How ready is your company for AI, and where do the first opportunities sit? Interviews with the people who run your processes, an AI-readiness scorecard, and a debrief with a first list of candidate use cases. The low-commitment way to find out whether a full Map is worth it.
Most chosen
AI Opportunity Map
Which AI ideas to build, and in what order. We zoom in on three to five priority processes, your own people surface the ideas, and I score every candidate on impact, effort, and data-readiness. For each one I check whether today's AI tools and your own systems can actually carry it. You leave with a short list of high-conviction, costed use cases, plus a clear list of what to ignore.
Fractional AI Lead
Your AI lead for a fixed term, on fixed days each week. I start with what the build needs: a phased roadmap, a deep data check on the top use cases, and a budget with build-or-buy calls. Then I lead the build with your team: architecture, review, the first working version, and the hand-over to your own people. No Map yet? Then the first two weeks are one.
AI Training
Hands-on training for your developers, in your own codebase. AI-assisted coding with Claude Code, Cursor, Codex, and GitHub Copilot. Building AI agents that can run in production. RAG, so a model answers from your documents instead of making things up.
See the trainingsAnd after the plan? The goal is that your company can build without me. By the end, the system runs on your side, with your people trained to keep it running. That is the point of the hand-over. I stay on as advisor for as long as that is useful.
How I think, on one page
This is an Opportunity Map for a fictional installer with 60 people. Eight ideas went in. Five made the short list and three got a no, each with a reason. Every engagement I do starts from a picture like this.
- Build now
- Build later
- Ignore
- Data not ready yet
The short list
- Rebuild the company website with AI-assisted coding
- AI drafts quotes from incoming requests
- AI reads supplier invoices straight into the ERP
- An AI assistant answers engineers' questions from the manuals
- Customer portal built with AI-assisted coding: orders, invoices, service history
Not now, and why
- Machine-learning forecast of parts demand
- AI chatbot on the website
- AI summaries of meetings and email
In the Opportunity Map
-
The map
Every candidate plotted on impact versus effort, flagged for data-readiness.
-
The short list
Three to five use cases, each with effort, rough cost, and the data it needs.
-
The no's, with reasons
The ideas you can stop discussing, and why.
-
The starting plan
What to do first, what comes after, and who needs to be involved.
-
The readout
A session with management and your tech lead where I defend every placement.
The Quickscan finds the first candidates for a map like this. The Opportunity Map is the full version. As Fractional AI Lead I build the short list with your team, and the AI Training teaches your developers to build the rest.
How an engagement runs
The same four steps in the Quickscan and the Opportunity Map. Only the depth differs. The Fractional AI Lead picks up where step four ends.
-
Explore
A kickoff workshop and short interviews with the people who know your processes best.
-
Select
A use-case session where your own team surfaces the ideas. No imported best practices.
-
Prioritise
Every candidate ranked and roughly costed. Enthusiasm alone does not get an idea onto the list.
-
Start
A readout with a concrete starting plan: what to do first, and what comes after.
A few questions clients ask
Isn't this just an expensive brainstorm?
A brainstorm gives you ideas. The Map gives you a prioritised, costed, data-checked plan, plus the no's. One avoided wrong bet pays for the whole engagement.
Why not use ChatGPT, or let our own team experiment?
Generic tools give generic ideas. They know nothing about your workflows or your data. I read your real processes, and I know under the hood what AI can and cannot do today.
What if our data turns out to be a mess?
That is a result, not a failure. It tells you exactly what to fix before you spend money on a use case that would have failed anyway.
We get the plan. Then what?
You own it, and your own people build it. If they need to get up to speed first, I train them. If you want the build led, I do that too, on fixed days a week. I stay on as advisor for as long as that is useful, and no longer.
Can you build it too?
Yes. As fractional AI lead I design it, build the first version with your team, and hand it over. For a build you need at least one developer on your side. That is who I hand the system to. After that I am your advisor, and the system is yours to run.
Where do you work, and in which language?
I am based in the Randstad, within an hour of Amsterdam, Rotterdam, The Hague, and Utrecht. I work with companies across the Netherlands, Dutch ones and international ones with a team here. On-site when that helps, remote when it doesn't. In Dutch or English, whichever your team works in.
Which AI opportunity is genuinely worth pursuing?
Book a short introductory call. We'll discuss where you stand, which questions are on the table, and whether one of these four is a sensible next step.
No obligation, practical, and no sales presentation.
