Process improvementmeetsSoftware Development
A vision for data-driven digital transformation
The challenge
The ambition.
Every company is busy digitalizing and automating. Making processes smarter, using data better, reducing manual work.
The reality.
But it often takes more time and money than expected. Getting people on board is hard. Projects overrun. And the result does not always match how the work is really done.
The consequence.
The big system choices do get made: a CRM, WMS or webshop. But the many smaller processes that fall in between stay stuck in Excel and workarounds.
The Pareto principle in action: it is exactly those processes that cost disproportionately much time, cause errors and keep experienced people from the work that really matters.
The cause.
The business knows exactly what is needed but speaks a different language than the IT people who build it. That translation step causes information loss and you notice it in the end result.
On top of that: for these processes there is no off-the-shelf software. They are too company-specific to build a SaaS product for and too small to justify a custom development project.
What if the person who optimizes the process also builds the software themselves?
That combination is worth gold.
The person who understands the process builds the solution themselves. No briefing to a developer, no translation step, no information loss. The solution fits because the builder knows the problem. For your organisation that means software that fits seamlessly with how the work is really done.
And that is possible now thanks to AI.
AI-assisted development makes it possible to build professional software yourself. Not tinkering but real applications, built with professional tooling. Those smaller processes that were left behind? Those are now opportunities. Think of a supplier portal to collect requests and evaluations in a structured way, a professional onboarding tool for new employees or a dashboard that brings scattered data together into one clear overview.
How it works.
The approach.
Using AI to build software is the new standard. People without a technical background experiment with AI, build something and call it software development. But without structure it is not a process. As a process improver, that is a familiar problem. That is why vibe-coding was treated as a process problem: the right tools and workflows were selected and turned into a full-fledged process of AI-driven software development.
The technique.
The central tool is Claude Code: an agentic development environment where the AI reads files, runs tests and makes decisions across a complete task. Not a single prompt, but a full working session.
Through the Model Context Protocol, Claude Code has access to external tools that semantically search existing production code on GitHub. That makes the difference. Where a standard LLM generates code from training data that may be outdated or incorrect, this approach finds working and tested implementations of the exact pattern that is needed. The resulting code is not generated but grounded in what works in production.
The process also has fixed quality gates. Automated checks for type safety, linting and test coverage run before every commit. Separate agents actively validate the codebase against current best practices and flag outdated patterns or security risks. The same principle as in any process improvement: quality is not a by-product but a built-in part of the way of working.
The result.
The result is a modern, well-secured and tested codebase, built on a process that scales. And what can be built? Almost anything. Custom applications, integrations, internal tools, dashboards, automated workflows and applications with built-in AI functionality. The combination of process improvement and AI-driven software development makes the possibilities bigger than they have ever been.
Curious what this could mean for your organisation?
Your own software suite. Tailor-made.
Process improvementmeetsSoftware Development.
Powered by Bram Schonenberg