AI product portfolio · Independent builds
Building AI products is easy to start. Knowing whether they work is harder.
I’m a product leader who builds working AI applications to understand the decisions behind them: evaluation, grounding, agent design, cost, trust and user experience.
What these projects demonstrate
These are experiments in product judgement, not claims of production deployment. Each project examines an architectural or product decision, measures behaviour and documents limitations.
Three recurring questions: How do we know the output is good? What should the model be allowed to do? How should the experience respond when the model is uncertain or wrong?
Selected builds
Code links are live destinations. Watch working demonstrations and explore the code. Some prototypes are still evolving; additional projects will be added as they are ready.