Case studies

A pipeline that analyses any company's trademark portfolio and surfaces unprotected marks, expired marks, renewal deadlines and geographic gaps. Raw registry data in, a ready-to-send sales brief out, in minutes per company.

Built to run across the whole target account list, every year, without anyone touching it.

One assistant that switches between teaching (trademark questions across dozens of jurisdictions, answers constrained to a sourced knowledge base) and selling (qualifying prospects with tact coded in).

One case per session, honest portfolio segmentation, even telling small portfolios they may not need the product.

A system that regenerates a multi-jurisdiction legal knowledge base every quarter, unattended, overnight, then notifies itself when done.

A small team moving like a large one, made literal: keeping the base current costs almost nothing.

A retrieval system over the EU AI Act and GDPR that returns exact source text with full provenance, no AI paraphrase, built for people who must cite the real wording.

Hybrid search (semantic and lexical, fused with RRF), evaluated against a labelled test set, with the architecture decisions and the failures documented.

A web platform that tracks and values European football cards across Topps and Panini, giving collectors a financial-market view of their collection. Listings and sales aggregated, every card identified by an LLM routing layer.

I'm the product owner and architect: I direct the build through AI agents, own every structural decision, and run the ops and the upstream data pipeline myself.

An MCP server that checks a candidate drug name against tens of thousands of medicines across EU and US markets, reserved WHO INN stems, and generic names, returning a risk-ranked verdict in seconds.

Three independent alert layers (brand, stem, INN) fused into one auditable answer. Built on official open data, runs entirely inside Claude in plain language.