Plan Assist cuts building licence assessment by 58 percent: AI enters British councils under human supervision
Google DeepMind announced the expansion of its smart tool Plan Assist to include 12 additional local councils in England, aiming to accelerate the review of building and renovation licence applications after early field results showed that the Gemini-based system could reduce the time planning officers spend evaluating homeowners' applications by up to 58 percent.
The field trial of the tool began in June within three local councils, Barnet, Camden and Durham, targeting the routine, labour-intensive tasks associated with ordinary building applications to free planning officials to focus on urban projects that require specialised expertise.
The results revealed that more than 70 percent of the technical reports generated by the tool received planning officers' approval without any amendment.including checking the request's compliance with local planning requirements and verifying that the mandatory documents submitted are complete.
The system is currently in a 12-month pilot phase using live applications in a real-world environment, after which the twelve new councils will be added in three successive batches, bringing the total number of participating bodies to fifteen, and if the tests are passed successfully the UK government plans to roll out the tool to all planning councils in England by 2027.
The experiment reinforces a decisive regulatory principle that requires every assessment to be reviewed and approved by a qualified planning officer before any official decision is issued.This pathway is linked to pressing economic calculations, as administrative bottlenecks in planning offices impose real costs that hinder the government target of building 1.5 million new homes by 2029.
This shift in British offices has practical implications for urban-planning and digital-government services sectors in the Gulf, Egypt and the Levant, where regional capitals have made great strides in digitising licence-application portals but the real bottleneck remains the manual review of plans against local building codes and municipal regulations.
The model provides decision-makers in the region with operational evidence that the time to issue residential licences can be cut from weeks to days without compromising structural safety, provided full legal responsibility remains with the certified engineer and the smart model is authorised for preliminary checks, thereby sparing governments the risk of property disputes.
At the labour-market level, this pathway reshapes the skills required of public-sector engineers and consulting firms in the region, as engineers move from routine procedural document checks to governing the outputs of intelligent systems and handling complex urban files, with growing demand for computing solutions that can accurately interpret building codes and local legislation.