Genomics plc
AI Strategy + Agent Engineering

// OBJECTIVE

Genomics plc uses polygenic risk score data to predict disease risk. They needed to automate a painful manual data processing pipeline and solve a long-standing biomedical ontology mapping problem.

// THE BREAKTHROUGH

We built an autonomous AI agent that reads research papers, interprets data structures, normalises datasets to internal formats, and decides when to mine for more data versus when to escalate to a human—with full action traces for audit.

// APPROACH

AI Action Plan and roadmap first, then rapid proof of concept to validate. For the ontology problem, we engineered a multi-stage retrieval and ranking system that maps between biomedical ontologies—including Genomics plc's proprietary one—where previous approaches had failed.

// OUTCOME

Data processing pipeline fully automated with human-in-the-loop escalation. Ontology classifier operational across their biomedical datasets.