UK tests AI to speed up routine planning applications
A pilot with three English councils classifies applications, summarises information and provides an initial assessment; the planning officer keeps the decision.
Applying for permission for a home extension or renovation can require councils to review many documents and references before reaching a decision. On June 17, 2026, the UK government unveiled an artificial intelligence tool to test whether it can ease some of that work in more routine cases, without turning planning permission into an automated decision.
The prototype is called Augmented Planning Decisions, or APD. According to the government, it is in early-stage testing with Barnet, Camden and Dorset councils. Its task is to triage applications, summarise relevant information and provide planning officers with an initial assessment that they can consider. The decision remains with the planning officer.
What the pilot is trying to change
The stated objective is to halve the average time for straightforward householder applications, from eight weeks to four. That objective concerns cases such as extensions or renovations, not replacing the review of complex projects. The government argues that reducing time spent on direct cases could allow teams to focus more on new housing, major developments and matters that require professional judgment.
The tool is being developed with Google DeepMind, Google Cloud and UK company Faculty. But the technology partnership does not change a basic public-administration principle: a recommendation produced by a system is not a decision. The digital team at the Ministry of Housing, Communities and Local Government stresses that every application must be assessed on its own merits and that the planning officer remains the key decision maker.
That boundary matters. Planning files contain more than structured data; they include local rules, technical documents, context considerations and observations from residents or other bodies. AI can help organise information or flag a missing document, but it does not remove the need to explain why an application is approved or refused.
Two tools, two problems
The UK announcement also presents Extract, a different tool that turns old planning documents and maps, including handwritten material, into usable data. Extract is already available to local authorities in England and addresses archive preparation; APD is aimed at the flow of new applications.
The distinction helps explain where AI may add value. First, digitise and structure information that is difficult to consult today. Then, support professionals when they review a file. Neither stage is the same as delegating planning policy to a model.
The government estimates that Extract can save manual transcription time after its trials. That is an estimate for a specific programme, not a guarantee that every council will achieve the same result. Likewise, APD still needs to show in the pilot that it speeds work without reducing the quality, consistency or transparency of decisions.
The relevant test is public
For a system of this kind to be useful, councils will need to know where every recommendation came from, what information it used and when it needs review. Residents, in turn, must retain clear ways to submit evidence, understand a decision and appeal it where appropriate.
The UK case does not present AI as a replacement for planners. It presents it as support for repetitive tasks, intended to return time to decisions that require expertise and public accountability. Its value will depend less on how striking the model appears than on pilot results: real timings, errors caught, recommendation quality and the ability of officers to challenge it.
Sources
Sources for this piece
This piece draws on 3 primary source(s), gathered during reporting.
This article was produced with artificial intelligence under human editorial oversight.
