Summary

• PlanAssist is expanding from three trial councils to 15 authorities for a 12-month evaluation.
• Initial tests reported a 58% reduction in time spent on selected planning tasks; officers retain responsibility for decisions.
• New delegation arrangements took effect on 31 October, independently of the AI trials.

The Ministry of Housing, Communities and Local Government is expanding trials of its PlanAssist AI system to 15 English councils after early testing found that it reduced the time spent on selected planning tasks by 58%.

The system brings planning applications, local policies, and site constraints into a shared working environment, helping officers locate relevant information and prepare initial assessments. Developed with Google DeepMind, Google Cloud, Faculty, and participating local authorities, it is intended to reduce routine administrative work while leaving planning decisions with qualified officers.

Camden, Barnet, and Dorset took part in the initial testing. Another 12 authorities will now join a programme running over the next 12 months, allowing the government to assess performance across councils with different planning workloads, local policies, and operational arrangements.

The expansion coincides with the introduction of the National Scheme of Delegation on 31 October. That reform changes which planning applications are determined by officers and which are referred to committees of elected councillors. The technology trial and the procedural changes have different functions, although the government expects both to contribute to faster decisions.

Preparing assessments from existing planning records

Planning officers must review several sources of information when assessing an application, including the proposed development, local planning policies, site restrictions, and other matters relevant to the decision. These records can be distributed across documents and systems that require substantial manual searching and comparison.

PlanAssist is intended to bring relevant information together and produce an initial assessment supported by identifiable evidence. Officers can use that work alongside their normal procedures, examining the underlying material before deciding how an application should be handled.

The distinction is important because identifying a policy or physical constraint does not necessarily determine the outcome of a planning application. Individual proposals can involve competing considerations, and officers must apply the relevant statutory framework and local policies to the particular circumstances.

Incorrect or incomplete source material could also affect an AI-generated assessment. An outdated planning document, for example, may contain policies superseded by later decisions, while a site constraint may require further interpretation before it can be applied appropriately.

The expanded trials will allow participating councils to examine both the time saved and the quality of the information provided. Because officers remain responsible for checking the resulting assessments, any operational improvement must account for the work required to review and correct them.

Early productivity findings face wider testing

The government reports that initial trials reduced the time officers spent on selected planning tasks by 58%. It has also described an ambition to reduce the processing time for routine householder applications from approximately eight weeks to four, using AI alongside other planning reforms.

These figures describe different measures. Time saved while preparing particular elements of an assessment is not equivalent to an independently demonstrated reduction in the total time taken to determine every application.

Overall processing times can be affected by the completeness of submissions, staffing, statutory consultation, site-specific issues, and the number of applications awaiting decisions. Improvements in document review may reduce one part of that workload without resolving every source of delay.

The government says householder applications account for almost 70% of annual planning applications, making them a substantial source of routine work for local authorities. Even comparatively small improvements in the time spent handling each application could therefore release officer capacity across a busy department.

The additional trial authorities are Ashford, Lambeth, Newcastle, Nottingham, Bristol, Cumberland, North Lincolnshire, Rochdale, Redbridge, South Gloucestershire, Stoke-on-Trent, and Westminster. Their participation will broaden the operating conditions under which the system is evaluated.

Differences between local planning records and procedures may prove significant. A system working reliably with one council’s policies and documents may need adaptation when introduced elsewhere, particularly where records are incomplete or stored in inconsistent formats.

Planning decisions change under separate rules

Under the National Scheme of Delegation, straightforward applications will increasingly be determined by planning officers rather than waiting for a meeting of the relevant planning committee. The government expects the arrangement to free elected members to concentrate on more significant developments.

Those changes affect the allocation of decision-making responsibilities, whereas PlanAssist addresses the administrative process supporting assessments. An application can benefit from improved access to information while still requiring consideration under the applicable delegation arrangements.

Delegation to officers is already part of local planning administration, but the national scheme is intended to create greater consistency in the way routine cases are handled. Local authorities must apply the relevant requirements rather than independently designing every aspect of their delegation arrangements.

The changes also raise questions about local democratic oversight. Elected councillors provide a form of accountability through planning committees, particularly where proposals attract strong local interest or involve significant policy judgments.

The Local Government Association has argued that the impact of the new arrangements should be reviewed earlier than the government’s proposed 2028 review. Its concerns relate to the balance between speed, consistency, and the opportunities for elected representatives to scrutinise decisions.

Because the delegation rules are already in force, councils are adapting their processes independently of the outcome of the AI trials. The operational effects of the two reforms will therefore need to be distinguished when assessing later changes in planning performance.

Better data supports wider deployment

The government intends to make AI planning tools available to every English council by the end of 2027, although it has explicitly stated that PlanAssist will not be mandatory software. Local authorities will be able to consider other technology suited to their requirements.

Alongside that objective, the government plans to require the digitisation and publication of important planning documents as open data by the end of 2027. More consistent digital records could support PlanAssist and other property technology systems by making relevant information easier to locate and interpret.

Digitising documents is not identical to improving their quality. Authorities will still need to maintain accurate records, reflect policy changes, and ensure that published material can be interpreted correctly by the people and systems using it.

The reforms also include investment in planning staff. The government says 420 planners have been recruited through supported schemes, exceeding its earlier commitment to support 300 additional professionals, while more than £14 million is allocated to relevant recruitment and training programmes during 2026/27.

Staffing and software address different pressures. Additional planners can increase the capacity to handle applications, while better information systems may allow existing employees to spend less time assembling documents and more time applying professional judgment.

The 15-council programme will test whether PlanAssist can produce reliable time savings across a broader range of authorities. Its results will provide a firmer basis for assessing the technology’s contribution to planning performance, separate from changes caused by staffing and delegation reforms.