Speeding up planning decisions with AI

Google DeepMind says its AI planning tool, PlanAssist, will be extended to 12 more local planning authorities in England after a pilot in Barnet, Camden and Dorset that began in June. According to the company, the tool cut the average time officers spent assessing householder applications by up to 58%, and officers approved more than 70% of its generated reports without any edits.

Built with Gemini and developed alongside the UK government, Google Cloud and Faculty, PlanAssist targets routine cases such as loft conversions, conservatories and extensions, which DeepMind says make up as much as 70% of officers' workloads. The system validates submitted documents and architectural drawings, checks proposals against heritage constraints like tree preservation orders and against local and national policy, looks for previously raised issues or objections, and drafts cited reports that officers can edit.

The project now enters a 12-month Beta phase on live applications, beginning with a subset of councils. The additional authorities will join in three cohorts, bringing the total to 15 across metropolitan and rural areas. If testing succeeds, the government plans to offer the tool to every council in England in 2027, with a qualified planning officer reviewing and approving each assessment before any decision is made.

The rollout matters because planning delays are seen as an obstacle to the government's target of 1.5 million new homes by 2029, and can leave households waiting months for minor approvals. The performance figures come from DeepMind's own account of early pilot testing.