Production AI systems we've designed and built.
A leading UK insurer needed to build their first customer-facing AI agent, but off-the-shelf RAG approaches weren't meeting the accuracy standards their regulators required. I led a team of three, designed a GraphRAG architecture with a custom ontology, built evaluation infrastructure, and took the system from concept to production.
Read case studyA fraud detection startup needed their AI to give consistent, deterministic results - their clients' trust depended on it. I designed a modular architecture with strict structured outputs, client-specific configurability, and evaluation frameworks that turned AI reliability from a subjective question into a measurable one.
Read case studyEnergy infrastructure professionals were drowning in thousands of pages of complex regulatory documentation. I co-founded the technical side, building a GraphRAG retrieval system and custom OCR pipeline that transformed unstructured legal text into AI-ready knowledge graphs. Concept to paying customers.
Read case study