Integration, automation, training
I advise and implement on integration and automation work, support Xurrent projects, and run training on software programming and architecture.
Services
Consulting and implementation wherever systems need to talk to each other reliably — from architecture through to day-to-day operation.
Consulting and implementation support around Xurrent — from rollout through to connecting it to the rest of the system landscape.
Training in programming and the architecture of software systems — hands-on, built around your actual project rather than textbook examples.
How I work
Programming languages have always made building software simpler. Instead of writing everything in machine code, you work in a higher-level language — an enormous leap. In today's world that simplification is gone. There are libraries that help manage the complexity of cloud-based development, but the underlying mechanics have to be built again and again.
One way out are platforms such as Jakarta EE or Spring. They take a great deal of distributed and concurrent complexity off your hands — but are complex to operate and cost real time to learn.
The other way are low-code and no-code platforms. They minimise the learning curve, yet usually break down on harder problems. Almost all of them can be extended with a full programming language — but that is exactly where the knowledge gap opens: straight from a drag-and-drop surface into the language, with nothing in between.
Both bother me in the same place: the effort has long since stopped being about the domain and moved to everything around it. That is what I work on in projects — and it is why I built a9script.
Business logic has to be deterministic. The same input must produce the same decision — today, tomorrow, and in an audit two years from now. AI systems are not: they are probabilistic. That is their strength, and precisely why they should not be the ones executing business rules.
The line belongs in the right place. AI may interpret, suggest and make sense of unstructured input. Execution belongs to deterministic code — repeatable, verifiable, traceable.
Causality is non-negotiable.
The tool for it
The automation and integration middleware this thinking went into: programs hold the business logic, the platform handles everything that makes modern cloud systems complicated. The product site has the detail.
Contact
The fastest start is a few sentences about the project, the timeframe, and where it currently hurts. Then I can contribute something useful in the first conversation.