DeepMind's PlanAssist: Gemini cuts UK planning review time 58%
Source
deepmind.google
Author
GoogleDeepMind
Date
Why it matters
A government deployment of a Gemini multimodalA model that works with more than text — reading images, audio, or video, and sometimes generating them too.Full definition → pipeline (document validation, policy checks, cited draft reports) reports a 58% time reduction and 70%+ no-edit approvals. It gives engineers real-world numbers for human-reviewed AI agentAn AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.Full definition → workflows in regulated settings.
Key takeaways · AI-distilled
PlanAssist has run in Barnet, Camden and Dorset councils since June; the more than 70% of its reports that officers approved without edits covered checks such as local-rule compliance and whether all required documents were submitted.
Before drafting a report, PlanAssist flags heritage constraints such as tree preservation orders and looks for issues or objections raised earlier, including by local residents.
Google DeepMind says routine householder applications make up as much as 70% of planning officers' workloads, and pitches faster handling of them as support for the UK goal of 1.5 million new homes by 2029.
If the beta succeeds, the UK government plans to offer PlanAssist to every council in England in 2027, with a qualified planning officer reviewing and approving each assessment before any decision.
Terms in this piece · Glossary
multimodal — A model that works with more than text — reading images, audio, or video, and sometimes generating them too.
AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
