Systems, built with AI
Internal tools, integrations, automation. Whatever the thing is that your people are doing by hand. The build method is the advantage: what used to take a quarter takes a couple of weeks.
Day one — building this in public
You have the lemons.
You have the lemons.
We make the juice.
I'm Mike. I build the systems companies actually need — compliance tooling, product data, internal automation — with AI as the build method. Working software in weeks, not slideware in quarters.
What I do
00
01
Getting unstuck
The project that stalled. The data nobody can untangle. The integration that was supposed to land last year.
02
Compliance systems
Product passports, GS1 resolvers, traceability data. The regulation has a date on it, so the software should be ready before it arrives.
Where to start
Most companies know something is wrong
before they know what to fix.
A session with your team is the cheapest way to find out which. We use your own processes as the material, not generic examples, and your people learn the tools on the work they actually do.
You leave with a team that can use it and a shortlist of what is worth building, in the order worth building it. If nothing on that list is worth building, I will tell you that.
- Format
- Half day or full day, on site
- Who
- The people doing the work, not only the managers
- Material
- Your processes, your data, your tools
- After
- A shortlist of what to automate first
Things I've built
GS1 Digital Link resolver
Product-passport resolver for a manufacturer's traceability pilot — spec-compliant, cloud-hosted, built against a hard regulatory deadline.
Brand-aware translation tool
Translation that holds a company's tone of voice across languages, instead of flattening it into generic machine output.
Transcription pipeline
Audio and video to transcript and summary, running on internal infrastructure. Shipped as a web tool and an iOS app.
Greenwashing checker
Scans marketing copy against sustainability-claim rules before it goes out, so legal review catches problems earlier and cheaper.
Sales & product tooling
Trade-fair lead capture with offline OCR, a product visualisation and engraving tool, and a dashboard measuring what all of it actually saves.
Day one
No portfolio site.
No case studies.
Not pretending otherwise.
Juiceworks started on 28 August 2026 with a domain, a camera, and a long career spent building systems inside a company that makes real things.
Most consultancies show you the polished end state. I'm showing the whole thing instead — what I charge, what I win, what I get wrong, and how the work actually gets built. If that's useful to watch, the channel starts at zero. If it's useful to hire, that's the same email either way.
- Based
- Netherlands, working anywhere
- Focus
- Compliance · product data · automation
- Method
- AI-assisted, shipped, documented
- Started
- 28 August 2026