On October 31, Google DeepMind said PlanAssist, a Gemini-built planning tool developed with the UK government, Google Cloud and Faculty, will be made available to 12 more local planning authorities in England, bringing the total to 15 for a 12-month Beta on live applications. In early pilots at Barnet, Camden and Dorset, DeepMind says the average officer time on householder assessment tasks fell by 58%, and more than 70% of the drafted reports were approved with no edits. Planning officers keep the final decision.
Key Takeaways
- ✓Scale: from 3 pilot councils to 15; the 12 new ones join in three cohorts for a 12-month Beta (DeepMind blog)
- ✓Efficiency: average officer time on householder assessment tasks fell 58% in pilots (vendor-reported; DeepMind's X post says "up to 58%")
- ✓Quality: more than 70% of AI-drafted reports were approved by officers with no edits (vendor-reported)
- ✓The UK government says, together with new rules, routine householder processing can halve from eight weeks to four, and it aims to offer the tool to every council in England in 2027 (GOV.UK)
Key Decision Metrics at a Glance
Turn your technical choice into a development budget
Compare 40 dev plans & simulate token costs vs $20/mo subscriptions
Project Links & Resources
Direct AccessIn-Depth Technical Analysis
Background and the pain point
The UK government wants 1.5 million new homes by 2029, but local planning approvals are badly backlogged. Householder applications such as loft conversions, extensions and conservatories are usually simple, yet they make up as much as 70% of planning officers' workload. Google DeepMind built PlanAssist with the UK government, Google Cloud and Faculty to give AI the first pass on this routine work.
How it works
PlanAssist is built with Gemini and handles an application in four steps. It validates the application documents and architectural drawings, checks the case against site constraints such as tree preservation orders and against local and national planning policies, looks up whether objections were raised before, for example by residents, and drafts a fully cited assessment report that officers edit before deciding. It is an assistant, not a decision-maker: a qualified officer reviews and approves every assessment.
Benchmarks and reported results
Pilots have run since June at Barnet, Camden and Dorset. DeepMind says average officer time on householder assessments fell 58% and more than 70% of reports were approved with no edits. The UK government says that, with new rules, routine processing can halve from eight weeks to four. These are official or vendor figures; no sample size, measurement method or independent evaluation has been published.
What it means for developers
This is a concrete example of an LLM entering a government workflow. The AI does not hand down a verdict; the pipeline is validate, check rules, retrieve history, draft with citations, then human sign-off. Teams building enterprise or public-sector agents can copy that shape, tie every conclusion to a checkable source, and use the no-edit approval rate to measure draft quality. PlanAssist is not sold commercially, and the 12-month Beta will decide whether it reaches every council in England in 2027.
Benchmark side-by-side against alternatives, or calculate monthly token cost vs subscription break-even.
Discussion & Comments
0Sign in to join the discussion
Connect with AI developers to exchange benchmark insights.