Google DeepMind tool cuts UK planning officer workload by 58%

Streamlining the UK planning process

The UK government is currently facing a significant challenge in meeting its goal of building 1.5 million new homes by 2029. A major hurdle in this objective is the administrative backlog within local planning authorities. Many simple applications for home improvements can take months to process, creating delays for families and the broader economy.

To address these bottlenecks, Google DeepMind has developed PlanAssist. This AI tool is designed to handle the heavy lifting of routine householder applications. By automating the most repetitive parts of the process, the technology allows planning officers to focus their expertise on more complex cases that require human judgment.

As Google DeepMind reported, the tool is part of a broader partnership with the UK government to explore how AI can transform public services. The goal is to use technology to unlock economic growth and speed up the approval of residential projects across the country.

How PlanAssist handles applications

PlanAssist functions by automating several specific steps in the application review process. First, it validates the submitted documents and architectural drawings to ensure all required information is present. This initial check ensures that the application can proceed without being sent back for missing paperwork.

The tool then moves into more complex automated checks against specific rules. It verifies the application against heritage constraints, such as tree preservation orders, as well as local and national planning policies. It also scans for any history of objections or issues raised by local residents in the past.

Finally, the system drafts a fully cited report based on these findings. These reports are then presented to planning officers, who can edit the text before making a final decision. This structure ensures that while the AI does the data gathering and drafting, the human remains the ultimate authority on every single application.

Proven results from pilot testing

The tool has already undergone pilot testing in the councils of Barnet, Camden, and Dorset. During these tests, PlanAssist demonstrated its ability to bring the average time spent on householder assessment tasks down by 58%. This is particularly significant because householder applications account for up to 70% of the total workload for planning officers.

The accuracy of the AI has also been a key metric for the developers. During the pilot phase, more than 70% of the reports generated by the tool required no edits from the officers. These reports included checks on whether an applicant submitted all required documentation and whether the project complied with local rules.

These results suggest that the AI is highly capable of handling the routine aspects of the job with minimal intervention. By providing high-quality drafts that are often ready for immediate approval, the tool allows the human workforce to manage a much higher volume of cases than previously possible.

Scaling to 12 more councils

Following the success of the initial pilot, PlanAssist is entering a 12-month Beta phase. This phase will involve live applications and will begin with a subset of councils. The project is expanding to include 12 additional local planning authorities across England in three distinct cohorts.

The selection of these new councils is intended to provide a diverse testing ground. The rollout will include both metropolitan and rural districts, allowing the developers to see how the tool performs in different community settings. This variety is essential for ensuring the AI works correctly regardless of local geography or specific regional requirements.

By the end of this expansion, a total of 15 councils will be using the tool. This testing phase is a necessary step before the government moves toward a wider deployment. The objective is to ensure the technology is reliable and effective across a wide range of different planning environments.

The road to a 2027 rollout

The long-term plan for PlanAssist is ambitious. If the current Beta phase and the expansion to 12 more councils are successful, the government intends to make the tool available to all councils in England by 2027. This would represent a major shift in how local governments manage property development and public services.

A key safety feature of this nationwide rollout is the requirement for human oversight. Every assessment generated by the AI will be reviewed and approved by a qualified planning officer before any final decision is reached. This ensures that the technology acts as a support system rather than a replacement for human expertise.

Ultimately, the goal of the partnership between Google DeepMind, the UK government, Google Cloud, and Faculty is to improve the lives of residents. By cutting backlogs and approving applications faster, the project aims to help families add nurseries, prepare homes for winter, and contribute to the UK's broader economic goals.