Councils across Britain are using artificial intelligence to scour their local areas for land that could be used to build homes, with one London borough uncovering more than 3,000 potential sites that planners had previously overlooked.
Results from a government pilot programme, published by the Ministry of Housing, Communities and Local Government, show how local authorities fed data into AI tools that identified parcels of land suitable for residential development. The findings come as Housing Secretary Angela Rayner renews her commitment to building 1.5 million new homes, a target she has described as challenging but non-negotiable.
The most striking result came from Lewisham in south London, where an AI tool called the Small Sites AI Finder identified 3,000 parcels of land with potential for housing. The council estimated that 9,747 homes could be built across these locations, which included infill plots, backland and side-street sites that traditional planning assessments had missed.
The tool, developed in partnership with north London architecture firm RCKa, used Ordnance Survey data to train an algorithm that could spot viable sites at a scale no human planning team could match. A single planning officer might assess a handful of sites per year. The AI processed an entire borough.
Reigate and Banstead council in Surrey offered a striking before-and-after comparison. Before the pilot, the housing team managed to assess just one parcel of land for development each year. Using a programme called Vestega, they created a longlist of 80 parcels, whittled that down to 19 viable sites and built a focused development pipeline that could deliver 400 homes.
The speed gains were consistent across the country. West Oxfordshire and Cotswold district councils reduced site assessment times from around ten weeks to one week. The Wirral assessed more than 800 sites during the pilot period, cutting assessment time by half. Durham County Council reported similar results.
Not every identification led to bricks in the ground, though. Basildon council in Essex used AI to identify 42 potential parcels, but only two have since received planning permission. Finding land is one thing. Getting consent to build on it is another entirely.
The government is now pushing further. A prototype AI system, built in partnership with Google DeepMind, Google Cloud and AI specialist Faculty, is being tested at Barnet, Camden and Dorset councils. It doesn’t find land. It triages planning applications.
The system summarises key information from applications and provides planning officers with an initial assessment. The claim is bold: processing times could fall from eight weeks to four. That would represent a meaningful change for anyone who has waited months for a decision on an extension or loft conversion.
A separate tool called Extract, which uses AI to convert old paper planning documents and maps into digital records, is already available to every council in England.
MHCLG has announced that 11 new council and tech company partnerships will take part in round six of the Proptech Innovation Fund, backed by £2.4 million of government money over eight months.
Liverpool City Region and West Midlands combined authorities will test Blocktype, described as an AI-powered platform for design-led housing capacity assessments across multiple sites. Birmingham and Reading will trial AI Gizmo for section 106 negotiations, the agreements through which councils require developers to fund local infrastructure such as schools, roads and GP surgeries.
Tower Hamlets and Barking and Dagenham in London will test PlanningHub software. The government expects to expand trials to up to ten additional councils later this year, with national rollout planned from 2027.
Norfolk and Suffolk planning departments are typically small teams working through complex constraints. North Norfolk sits within an Area of Outstanding Natural Beauty. Coastal areas face flood risk assessments. Conservation areas in towns like Holt and Burnham Market add layers of complexity that slow down even straightforward applications.
AI tools that can triage applications, digitise legacy records and identify overlooked infill sites could be transformative for councils struggling with both housing targets and limited resources. The East of England faces persistent pressure to deliver new homes while protecting the character that makes its market towns attractive in the first place.
Data from The Ivybridge Collection’s property market reports across 324 locations shows that housing supply constraints continue to underpin prices in desirable areas like Norwich and Southwold. Anything that accelerates the planning pipeline, even modestly, has implications for both buyers and sellers.
Basildon’s experience is a useful corrective. Its AI found 42 potential sites across a borough where two-thirds of the land is classified as Metropolitan Green Belt. Two received planning permission. Sevenoaks in Kent, where 93% of land sits within the green belt, faces similar constraints that no algorithm can override.
AI can identify land. It can speed up paperwork. It can digitise decades of planning records that currently sit in filing cabinets. What it cannot do is resolve the political and legal battles over where homes should actually go. Section 106 negotiations, green belt protections, conservation designations and public opposition remain stubbornly human problems.
The shift is real, though, even if the headlines about “AI snooping” overstate the drama. East Herts District Council used the pilot to open a digital call for sites, receiving 290 submissions from estate agents, landowners and residents. That kind of public engagement, conducted through technology rather than parish hall meetings, suggests planning is changing in ways that go beyond back-office efficiency.
For property owners across Norfolk and Suffolk, the practical effects may take years to materialise. National rollout isn’t expected until 2027. But councils that adopt these tools early will have a clearer picture of their developable land, faster processing times and, potentially, more homes reaching the market. In a region where the gap between housing demand and supply continues to shape prices, that matters.
