Built by Willem Gielen
Cardiologist, physician-builder, and AI expert
Lenormand Intelligence is my independent project at the intersection of traditional Lenormand, structured reasoning, and modern artificial intelligence.
Why I built this
I'm Willem Gielen, a cardiologist with a longstanding interest in technology, clinical decision-making, and artificial intelligence. In medicine, I work in a world where complex information has to be reduced to something useful: identify the important signals, understand uncertainty, and communicate a conclusion clearly.
Lenormand Intelligence is not a medical tool and it does not turn card readings into medical, financial, or legal advice. Lenormand is a separate symbolic practice. What makes it interesting to me is the engineering problem: can a system represent a question, apply a defined reading method, and produce a useful answer without pretending that an unconstrained language model is an oracle?
That makes this project a practical experiment in structured reasoning and AI engineering. The cards provide a compact domain; the software provides the discipline needed to make its reasoning inspectable and repeatable.
The engineering approach
Deterministic logic first
Spread layouts, card relationships, timing evidence, and question framing are represented in code rather than left to improvisation. The same input should lead to the same structural context.
Constrained AI
AI is used for language and synthesis inside a defined context. Structured output, explicit prompts, and limited responsibilities keep the model from replacing the method.
Validation over confidence
Generated content is parsed, normalized, and checked before it is rendered. When an output does not satisfy the contract, the system should recover safely or fail clearly.
Honest boundaries
A reading is an interpretive tool for reflection, not a prediction engine or professional service. The site should make that distinction clear rather than oversell certainty.
A work in progress
Lenormand Intelligence is intentionally an evolving project. I use it to test how domain models, deterministic software, structured output, and language models can work together without hiding the limits of any one of them.
The goal is not to make a machine sound mystical. It is to build a clear, useful, and technically honest interface for exploring a traditional system of symbols.
Willem Gielen