Descriptions:
Eoin Mulgrew, who leads cross-government AI transformation at the Number 10 Downing Street data science team (10DS), presented at the AI Engineer conference on how a small embedded engineering team is driving AI adoption across the UK public sector. The context is stark: 7.25 million people on NHS waiting lists, 350,000 backlogged court cases, and a Tony Blair Institute estimate of £40 billion in annual productivity gains available through AI in government.
The operational model Mulgrew describes is a fellowship program that places senior AI engineers directly inside high-priority government departments for three-to-six-month rotations, building durable in-house capability rather than handing work to outside consultants. Concrete outcomes include an AI tool for analyzing the entire UK statute book — replacing a planned £1.5 million external legal contract and, critically, enabling continuous analysis that keeps pace with new legislation rather than requiring periodic expensive repeats. A second tool, a delivery red-teaming system now used daily inside Number 10, interrogates incoming project reports and flags departments with systematic optimism bias or habitually ineffective risk mitigations.
Mulgrew is candid about structural barriers — civil service pay scales, hierarchy, and the genuine difficulty of retaining high-performing technical talent in government. He also points to recent public-facing AI dashboards tracking the UK’s AI Opportunities Action Plan rollout as evidence that transparency and accountability are improving alongside delivery capacity. An upcoming public service affecting millions of UK residents is mentioned but not detailed, embargoed at the time of the talk.
📺 Source: AI Engineer · Published May 18, 2026
🏷️ Format: Keynote Launch
