Building a council's AI capability, from strategy to a production agent factory
Set and delivered the AI strategy for a council of around 4,500 staff – the rollout approach, the internal capability to sustain it, and a productionised agent factory model that turns validated use cases into repeatable delivery.
- AI strategy and rollout approach established from scratch, then delivered under cabinet-approved governance
- Internal capability built so the council can run and extend AI delivery without external dependency
- Agent factory model productionising validated use cases into repeatable releases
- Governance and sign-off framework spanning data, technical and risk review at every release
Local government does not get to move fast and break things. Every release has to survive procurement law, data protection review and a council chamber's worth of scrutiny. The brief was not simply to move 4,500 people onto new tools – it was to decide what the council's AI approach should be, build the capability to run it, and hold on to risk, cost and long-term ownership while doing so. Two streams ran at once: a broad adoption program covering communications, enablement and the unglamorous infrastructure that makes "just use Copilot" actually true, and a narrower agentic delivery track validating specific use cases through to live operation. Those validated cases are now consolidating into an agent factory model – a repeatable way to build and release agents that make council processes more efficient and get residents better services, rather than a series of one-off pilots.