source: google deepmind: unlocking uk house-building with ai-accelerated planning
level: business
the uk government wants to build 1.5 million homes by 2029, but local planning authorities face paperwork backlogs. a new ai prototype built with gemini, co-developed by google deepmind, google cloud, faculty, and councils in barnet, dorset, and camden, acts as an assistant for planning officers. it handles data extraction and case analysis to cut decision times for householder applications like loft conversions and extensions.
the tool consolidates data from backlogs, highlights missing information, and extracts site details onto one screen. it identifies relevant national and local policies, pre-assesses compliance, and provides citations for officers to verify. it summarizes public consultation feedback, picking out key objections or precedents. it also drafts the final report, including rationale and proposed conditions, while the officer stays in full control as the decision-maker, reviewing and editing every line.
this builds on extract, a tool already available to all english councils that turns legacy planning documents into usable data in minutes, saving around 255 hours of manual work per council each year. the new prototype is expected to be available nationally from 2027. by reducing administrative work, it aims to free up planners for complex cases and speed up home improvements, supporting economic activity and housing targets.
why it matters: it shows how ai can streamline government processes, directly reducing bureaucratic delays in housing and offering a model for public service delivery worldwide.
source: google deepmind: unlocking uk house-building with ai-accelerated planning