Not everything has to be AI.

Pragmatic AI development. AI is one of my work tools; it only ends up in your product where it really adds value.
3 - 5clients at a timereal attention, no ticket stack
from €295per month3-month minimum, 1-month notice
25 yearsback-end experiencein production since 2001

Everyone seems to want AI in their product these days. Not me, or at least not just like that. I build custom software where AI mostly lives in the development process, and in your end product only where it truly adds value. The rest is plain code; faster, cheaper, and without depending on an outside model provider.

problem.md

"AI in everything" sounds smart, the bill comes later. Most implementations I run into hit the same four problems:

  • Monthly API costs that grow with your usage; sometimes hundreds of euros for features plain code can handle too.
  • Provider lock-in with vendors that raise their prices, retire models or suddenly change their policies.
  • Software that feels slow because every action takes a round trip to an external API.
  • Customer data going to third parties when the feature doesn't strictly need it.

The AI-everywhere approach is good for vendors' marketing; not always for the wallet of a small business. Many features now sold as AI work fine with traditional code: more reliable, faster, and without a monthly bill. The trick is knowing when AI really adds value, and when it doesn't.

approach.md

AI on my workbench

I use AI every day to reason, build and test faster. You get the time savings back in price and lead time; not the dependency in your end product.

Think first, integrate later

For every feature, one question: can plain code do this? Often: yes. Sometimes AI is the right call; then we build it smartly, with fallbacks for when the provider is down.

You stay in control

No black boxes, no vendor lock-in. Code, data and choices are yours; even if you want to continue with someone else tomorrow.

The result is software that's fast and runs reliably, with predictable costs and without your data leaving the building.

packages.json

Basic

Keep what's there running

€295/month

Meeting
quarterly meeting
  • Monitoring, updates and bug fixes
  • Small maintenance on existing integrations
  • Monthly report
  • Quarterly conversation about direction and priorities
// most popular

Partner

Build on what's there

€595/month

Meeting
monthly
  • Everything in Basic
  • Ongoing development of features and integrations
  • Monthly meeting with roadmap
  • One concrete improvement per month
  • Proactive suggestions from the AI landscape

Strategist

AI strategy + building on what's there

€1,195/month

Meeting
every two weeks
  • Everything in Partner
  • Strategic input on AI use and architecture
  • Strategy meeting every two weeks
  • Quarterly AI architecture review
  • Roadmap document updated twice a year
  • Setup fee: first month at 2x the monthly price
  • Minimum term: 3 months, then 1 month notice period
  • Out-of-scope work: €110/hour
  • You can switch tier every month
  • Hosting, API tokens and external SaaS go straight through your own account; I help with choosing, setting up and monitoring. Re-billing is possible on request.

process.json

Not one-off projects, but a fixed amount per month for ongoing development, maintenance and advice. Software is never "done"; a delivered project starts aging from day one. A subscription forces me to deliver something worthwhile every month.

  1. Introductory call

    A short conversation to see whether your question fits my approach. No sales pressure; if I think you're better off with someone else, I'll say so.

  2. Setup month

    I charge 2x the monthly price for the first month. That covers getting up to speed on your stack, reading your codebase and setting up monitoring.

  3. Monthly cadence

    Monthly report, one concrete improvement, a status update on what's happening in AI land, and proactive suggestions. No hour banking; you don't buy 7 hours, you buy a partner.

  4. Adjusting course

    Every month you can switch tier. The work grows, shrinks or shifts; the price adapts to it.

The minimum term is 3 months, then a 1-month notice period. Work outside the scope is billed hourly at €110.

benefits.json

Lower running costs

No €X a month in API tokens for functionality plain code can handle too.

No lock-in

A price hike at OpenAI or Anthropic doesn't just hit your operating costs; with me, AI only goes where it's strategically needed.

More reliable

Local code keeps running when an external API is down. Where AI is involved, it comes with fallbacks.

Faster

Code running on your own server is faster than a round trip to an external API.

Privacy

Your customers' data stays with you and doesn't go to a third party unnecessarily.

Personal attention

At most five clients at a time; no account managers, no ticket systems, just a direct line to the builder.

More sustainable, as a bonus

Fewer trips to AI data centers means less computing power. Not saving the world, but an honest side effect.

faq.json

Why a subscription, not a project?

I do one-off projects too; for a defined job with a clear end, a project price is fine.

But software is rarely truly done, and a project price invites cutting quality once the hours run out. A subscription invites delivering something worthwhile every month. Where that works out better for you, I prefer it.

Why I work this way, and why 'done' doesn't exist, you can read in this article.

What if I don't have 7 hours of work a month?

Then you're on Basic (3 hours), or we switch to a lower tier for a month. Switching tiers is possible every month.

You don't pay for hours you don't use; you pay for the partnership.

Why 2x the monthly price in the first month?

A new client costs me more time than an existing one; going through the codebase, getting to know people, setting up monitoring. The setup fee covers that.

After that, I'm in your flow.

What do you do when you run into out-of-scope work?

Then I stop and talk to you before I continue. A subscription is an agreement; I don't quietly absorb work that falls outside it.

Usually we pick one of three: move up a tier for a month, add it on an hourly basis (€110/hour), or push it to next month. You hear it upfront, not on the invoice.

How many clients do you take on at once?

At most five, and at Strategist level at most three.

That scale fits personal attention; any more and I'd be betraying my own proposition.

Do you really not build AI into my product?

I do, if it truly adds value.

Classifying incoming email, a summarizer for long documents, a question-and-answer layer on top of your knowledge base; those are legitimate cases.

But I won't put an LLM in a form validation or a price calculation; plain code does those better.

What about the AI Act?

Since February 2025, a company that uses AI has to make sure the people working with it are AI-literate. That's in Article 4 of the EU AI Act.

I've completed the AI Awareness training at Proud Nerds myself: all four modules, every test flawless. The framework is ISO/IEC 42001 and that same Article 4. You'll find it on my CV.

In my work, the human has the final say. AI generates, but doesn't decide; what that looks like in practice, you can read in AI in your development process, not in your product.

If I build AI into your product, we first decide together on the impact and the risk class, and on who is the provider and who is the deployer. That's not legal advice; for that, a lawyer remains the right address.

What if I temporarily have no budget?

Then we talk. By agreement we can pause for one or two months without ending the collaboration.

I'd rather have a short pause than lose you altogether; I'm building your system for the long term, not for a quarterly figure.

Are LLM tokens and external SaaS included in the subscription?

No, and that's deliberate.

LLM tokens (OpenAI, Anthropic) and external SaaS scale with usage and vary too much to fit honestly into a fixed monthly price; you pay the vendors directly, on your own account, so you can always see and steer the usage yourself.

What I do within the subscription: pick the right stack, set everything up and flag every month whether you can save or scale up.

What about hosting; do you handle that too?

By default you buy hosting, domain and SSL on your own account. Set up once, at cost, and you stay the owner; no lock-in on me for access or migration.

Prefer to keep it with one party? That works; I'll arrange a VPS and take over the management (updates, monitoring, backups).

I bill that separately from your regular subscription, so the time in that subscription stays reserved for strategic development and not for server maintenance.

Do you also work with clients outside small and medium-sized businesses?

My approach fits best with small and medium-sized businesses and self-employed people who value direct collaboration and personal attention.

For large organizations with complex procurement processes, vendor rosters and long tender trajectories, I'm often not the right choice; I'll say so honestly in the introductory call.

contact.json