The UK government is expanding its partnership with Google DeepMind to bring artificial intelligence into local planning services, introducing PlanAssist, an AI-powered tool designed to help councils process routine housing applications faster. Early testing has shown that the technology can reduce the time planning officers spend assessing applications by up to 58%.

The expansion marks a new phase for PlanAssist, which was developed by Google DeepMind in collaboration with the UK government, Google Cloud, AI company Faculty, and local authorities. Following initial testing across Barnet, Camden, and Dorset, the tool is moving into a 12-month beta phase that will bring 12 additional councils into the programme.

The initiative aims to address one of the persistent challenges facing England’s planning system: the time it takes to process relatively straightforward applications for home improvements, including loft conversions, conservatories, and house extensions.

According to Google DeepMind, routine householder applications can account for as much as 70% of planning officers’ workloads. Although many of these applications involve relatively straightforward assessments, reviewing documentation, checking planning policies, and identifying relevant restrictions can consume considerable staff time.

PlanAssist is designed to take much of that administrative work off planning officers’ desks.

How Google’s PlanAssist AI works

Built with Google’s Gemini technology, PlanAssist brings together the information planning officers need to assess an application in one place. Rather than requiring officers to manually work through multiple documents and planning records, the system automates several preliminary assessment tasks.

It begins by checking application documents and architectural drawings to establish whether the required information has been submitted. It then examines relevant planning restrictions, including heritage considerations and tree preservation orders, alongside applicable local and national planning policies.

The system can also identify previously raised concerns or objections, including feedback from local residents. Using this information, PlanAssist prepares an evidence-backed assessment report with citations that planning officers can review and edit.

This approach is intended to reduce the time spent gathering information and completing routine checks, allowing professionals to concentrate on more complicated aspects of planning applications.

Importantly, PlanAssist does not independently approve or reject applications. Qualified planning officers remain responsible for reviewing the AI-generated assessments and making the final decisions.

Early testing shows a 58% reduction in assessment time

The initial pilot programme began in June 2026, with Barnet, Camden, and Dorset councils participating in the development and testing process.

Google DeepMind reports that PlanAssist reduced the average time planning officers spent on householder assessment tasks by 58%. More than 70% of the reports generated by the system were also approved by officers without requiring edits.

These reports covered routine assessment requirements, such as checking whether an application complied with local planning rules and whether applicants had submitted the necessary documentation.

The results offer an early indication of how AI could help local authorities manage administrative workloads. However, the technology is now moving into a broader testing phase involving live applications, which will provide a more substantial opportunity to evaluate its performance across different planning authorities.

12 more councils join the 12-month beta programme

The next phase will expand PlanAssist from its original three pilot councils to 15 local planning authorities across England.

The 12 additional councils joining the programme are Ashford, Lambeth, Newcastle, Nottingham, Bristol, Cumberland, North Lincolnshire, Rochdale, Redbridge, South Gloucestershire, Stoke-on-Trent, and Westminster.

The authorities will be onboarded in three cohorts during the 12-month beta programme. Their participation will help evaluate how the technology performs across different communities, including metropolitan and rural areas.

If testing is successful, the UK government intends to make PlanAssist available to every council in England in 2027. The expansion is part of a broader effort to modernise planning services and support the government’s target of building 1.5 million new homes by 2029.

The government is also introducing changes to planning procedures. From October 31, 2026, straightforward applications will be decided by expert planning officers rather than waiting for a planning committee meeting. This is intended to reduce unnecessary delays for homeowners and small businesses.

Alongside these changes, the government says it has recruited 420 additional planners through government-backed programmes, exceeding its earlier commitment to recruit 300.

AI supports planning officers rather than replacing them

PlanAssist represents another example of how Google DeepMind is bringing its AI technology into public services beyond conventional consumer applications.

For homeowners, the potential benefit is straightforward: routine applications for extensions, conservatories, and other improvements could move through the planning process more quickly. For councils, automating repetitive checks could free up valuable time for applications involving more complicated planning considerations.

However, the expansion into live applications will be an important test. The early results are promising, but councils will need to evaluate the system’s accuracy, consistency, and usefulness across different local planning requirements.

Google DeepMind says every assessment will continue to be reviewed and approved by a qualified planning officer before a decision is made.

If the wider beta programme delivers results similar to the initial pilots, PlanAssist could become an important part of England’s efforts to modernise its planning system, reducing administrative work while keeping professional judgement at the centre of housing decisions.