Planning AI could be available to every council in England by the end of 2027 under a new commitment in the UK Budget published on 28 October 2026. Ministers have also promised to require councils to publish key planning documents as open data.
The Budget names a tool called PlanAssist and says wider access will depend on further testing. The government wants the technology to reduce the time needed for routine planning decisions, but it has not announced that applications will be decided automatically or that the new system is ready to deploy.
For applicants, agents and planning officers, the bigger issue may be the data sitting behind the technology. Better access to local policies, constraints and previous planning records could make applications easier to assess. But the new deadline is a policy commitment, not evidence that every council already has reliable records or working artificial intelligence software.
What the Budget says about planning AI
The official Budget 2026 document commits the Ministry of Housing, Communities and Local Government (MHCLG) to make PlanAssist available to councils in England by the end of 2027. That offer is expressly subject to further testing
. Ministers also say the ambition is to halve the time taken to deal with routine applications.
A second commitment requires local planning authorities to digitise and publish key planning documents as open data by the end of 2027. The Budget describes this as removing obstacles to wider adoption of artificial intelligence in planning. It does not, in that passage, specify every document category, the proposed legal instrument, an enforcement mechanism or a staged council-by-council implementation programme.
That distinction matters. Planning AI is a promised service and data reform, not a new power to ignore planning policy or bypass ordinary application procedures. Nor does the 2027 target mean the statutory determination periods have already been cut in half. Further technical detail and testing results are still required before those practical claims can be measured.
How this differs from the earlier planning AI pilot
In June, MHCLG described an early-stage artificial intelligence prototype being tested with Barnet, Camden and Dorset councils. The earlier announcement called it an Augmented Planning Decisions tool, designed to help officers triage applications, find relevant policy and constraints, and organise information for their assessment. It aimed to reduce the processing time for suitable householder cases from eight weeks to four.
The department’s June explanation said the initial development involved government teams and partners including Google DeepMind and Faculty. The published Budget uses the name PlanAssist. It does not explicitly establish in its text whether PlanAssist is exactly the earlier prototype, a renamed version or a subsequent product. That should be clarified rather than assumed.
Another tool, Extract, was already made available to English councils in June. It is designed to turn older planning records, including scanned documents and maps, into usable data. Extract and the decision-support prototype serve different functions. Calling everything planning AI risks overlooking the importance of preparing accurate underlying records before software can analyse them.
Open planning data could be the more lasting change
For the planning system, reliable and accessible information is vital. An officer assessing an extension or change of use needs to know which policies apply, whether there is a conservation area or Article 4 Direction, and whether earlier decisions reveal relevant site history. Poorly indexed paper plans and inconsistent electronic records slow down that work even without any involvement from artificial intelligence.
The government has previously identified national planning datasets covering conservation areas, Article 4 Directions and tree preservation orders. A separate Digital Planning Improvement Fund page lists those datasets and listed buildings among the records that participating authorities must publish. That fund has its own conditions and application timetable; the Budget does not say that every authority receives funding automatically.
For planning AI to be dependable, a published dataset needs more than a filename. Boundaries must be accurate, adoption and confirmation dates must be correct, superseded directions must be identified, and current policy should be distinguishable from draft policy. An incorrect Article 4 boundary could mislead an applicant into thinking permitted development rights exist where they do not.
This is particularly relevant to local restrictions explained in Planning Geek’s Article 4 guide. Digitising a direction is not the same as checking that it remains in force for the relevant property and development type.
Planning AI will not remove professional judgement
In its June account, MHCLG was explicit: A human planning officer will always be involved in the decision-making process.
It said the technology should assist with information gathering and routine administration, while officers remain responsible for weighing the planning merits of individual cases.
That is an important safeguard. A policy search may identify the correct paragraph yet still fail to understand the physical relationship between two houses, the significance of a heritage asset or the weight to give a recent appeal decision. The facts supplied to the system must be checked, and the final officer reasoning needs to be intelligible to the applicant and the public.
There is also a difference between officer support and committee reform. England’s new national scheme of delegation changes which applications are decided by officers and which may go to committee. It does not authorise a computer system to make planning decisions instead of a legally accountable decision-maker. Our guide to the national scheme of delegation explains that separate statutory framework.
For planning AI to improve confidence rather than simply increase throughput, councils will need to show how information is checked, how mistakes are corrected and who takes responsibility for the resulting decision. Faster paperwork is useful; a faster but poorly reasoned refusal is not.
What councils and applicants should watch next
The next test is practical delivery. The June prototype was still being evaluated with a small group of councils. The Budget now sets a national ambition for the end of 2027, but does not publish a final timetable for every authority, the technical specification of PlanAssist or independently verified results from a completed nationwide trial.
Authorities should watch for formal data standards, instructions on which planning documents must be made openly available, procurement and security requirements, and any new statutory instrument that puts the data obligation into effect. Applicants should watch for clearer access to up-to-date constraints and transparent reasons for decisions rather than assuming the next application will be determined in four weeks.
The Digital Planning Improvement Fund currently indicates that eligible English authorities can apply from 12 November to 11 December 2026 for up to £50,000 to improve data and digital capabilities. That is a separate programme, and the dates and eligibility should be checked directly with the scheme before any authority relies on the money.
The geographical limit is also important. These particular Budget commitments concern councils in England. Planning is devolved, so they should not be reported as an automatic 2027 rollout in Wales, Scotland or Northern Ireland.
I would welcome better planning records that anyone can inspect and analyse. But I would judge planning AI on verifiable improvements in decision quality, transparency and actual elapsed time, not simply the number of councils offered a software login. The Budget sets a direction. Whether the technology and the underlying data can deliver is the next story.
Original government sources
Read the Budget 2026 HTML document, the June announcement of AI planning pilots, the MHCLG digital planning explanation and the Digital Planning Improvement Fund guidance.
0 Comments