Dorset Council trials AI agents for planning applications
The UK government and Google DeepMind are trialling an AI prototype with Barnet, Camden and Dorset councils to halve planning application processing times from eight to four weeks. Developed under an MHCLG contract with Google Cloud, Google DeepMind, and UK AI firm Faculty, the tool triages householder applications, summarises key information, and gives planning officers an initial assessment for review. GOV.UK reports alpha trials began in May 2026 across three councils, targeting the 70% of England's approximately 350,000 annual planning submissions that are householder applications. A separate AI tool, Extract, is now available to all local planning authorities to convert decades of paper planning records into structured digital data. If trials succeed, national rollout is planned for 2027, with every assessment reviewed and approved by a qualified planning officer before decisions are made.
What happened
The UK government announced two new AI tools for England's planning system. A prototype AI agent, built under an MHCLG contract with Google Cloud, Google DeepMind, and UK AI firm Faculty, is in alpha testing with Barnet, Camden, and Dorset councils. GOV.UK reports the prototype triages householder planning applications, summarises key information, and provides planning officers with an initial assessment. The government says the target is to halve average processing times from eight to four weeks. A second tool, Extract, developed by the government's Incubator for AI (i.AI) with MHCLG's Digital Planning programme, is now freely available to all local planning authorities in England to convert decades of paper planning documents into structured digital data.
Partnership and funding
GOV.UK states MHCLG is funding the Augmented Planning Decisions (APD) prototype with a GBP 8.2 million contract awarded to Google Cloud, Google DeepMind, and delivery partner Faculty. Alpha trials started May 2026. Marc Waner, CEO of Faculty, said the tool gives planning officers clear recommendations - with humans retaining final sign off. Lila Ibrahim, Chief AI Readiness Officer at Google DeepMind, said the partnership aims to significantly cut decision times, freeing up planners to focus on the future.
Scale and context
Householder applications account for nearly 70% of England's roughly 350,000 annual planning submissions. GOV.UK notes that Extract is expected to save the average council around 255 hours of manual document review work per year, down from over 500 hours before the tool. If the APD prototype passes trials, the government plans to expand to up to 10 additional councils later in 2026, with national rollout from 2027. Every assessment will be reviewed and approved by a qualified planning officer before any decision is made.
Editorial context
Public-sector AI agent pilots that scope narrowly to document triage, summarisation, confidence scoring, and human-in-the-loop review represent a pattern now visible across multiple governments. For practitioners, deployments of this type emphasise data quality for unstructured documents, robust record linkage, auditable decision trails, and UI design that surfaces model uncertainty to human reviewers. The consortium model here - Google DeepMind (core AI), Google Cloud (infrastructure), Faculty (delivery and public-sector integration) - reflects an emerging structure for government AI contracts in the UK.
Scoring Rationale
A government AI pilot involving Google DeepMind and Faculty under a named GBP 8.2M contract, with a defined 2027 national rollout timeline, is notable for practitioners tracking public-sector agentic AI deployments. The consortium structure (DeepMind + Faculty + Google Cloud) and the human-final-review architecture make this a practical reference case, though it remains in alpha with no production results yet.
Practice interview problems based on real data
1,500+ SQL & Python problems across 15 industry datasets — the exact type of data you work with.
Try 250 free problems
