Recent work.
Four projects that show what translating, building and making AI work look like in practice. Most of the work is in agri and food. The same logic applies wherever bulk is counted, weighed, stored and transhipped.
Trading platform for agricultural bulk
A platform that knows what a batch is.
A trading company in agricultural bulk goods is bringing its entire operation onto one platform on Microsoft Dynamics 365. The trade is unforgiving: weights change in transit, a grading determines the price, and the same relation is a supplier today and a customer tomorrow.
I work here as product owner and solution architect on the client side. I translate the trade into requirements, manage the vendor and make sure the system follows the trade rather than the other way around. Over 1,200 work items, from requirement to acceptance test. I also lead the AI track: AI Custom Vision that grades a batch from a photo.
Translating: the trade in the requirements, and AI where it actually measures something.
Semantic layer for a food platform
Data AI tools can trust.
A SaaS platform for food processing wants AI in its own product. The problem: raw production data means nothing to a model. A tag number at a timestamp is not an answer, and without context any AI layer on top becomes guesswork.
I developed the AI strategy: a semantic layer on top of the production data, with a single gateway through which AI tools work safely and in a controlled way. The strategy has been adopted.
Making AI work: structure first, so AI gives answers that are right.
AI strategy for a technical company
Starting at the start.
A technical company in industry wants to get going with AI. The question is not which tool, but where to begin and how to set up the organisation for it.
I advised the management team on exactly that: where AI pays off first, what needs to be in place in data and processes, and how to assign ownership and ground rules so it doesn't remain a set of loose experiments.
Making AI work: direction first, then technology.
Building in-house
Built in-house, in production.
Alongside client work, we build software of our own. ViQ Planner (viqplanner.nl): our own planning application, in production. And our own algorithms for recognising structure in industrial production data and validating model output.
Why that matters here: someone who builds advises differently. We know what an integration costs, why a data model breaks and when a demo is not yet a product.
Building: proven with our own work.
"Robin brings immediate clarity and decisiveness to digital transformation projects. With his strategic insight and pragmatic approach, he consistently delivers measurable results."