Disruption and innovation: Scaling change in UK planning and construction with AI

AI is already reshaping planning and construction. The question is whether we engage proactively, or let an inability to adapt cost us control of our future.

Last week at UK Construction Week 2025, a session by Architect George Clarke, and Jacqueline Glass, Dean, The Bartlett, UCL’s Faculty of Built Environment, the real issues were laid bare.

[edit] What’s happening

There is a quiet revolution in planning workflows. Local planning authorities (LPAs) are beginning to analyse thousands of consultation responses and objections using AI, not to replace planners, but to identify bottlenecks and understand community concerns at scale. These tools work with real local data, drawing on word patterns and associations in representations, officer reports and evidence bases. The outcome is faster triage, clearer maps of issues, and better-informed human decisions.

Milton Keynes offers a grounded example of both the disruption and the benefits. In a recent Linkedin post, Jon Palmer, Head of Planning at Milton Keynes City Council, reflected on the team’s digital journey, including the hard parts and how AI is beginning to help:

“Twelve months ago we were about to switch off our old system. It was not pain free and a backlog built up which resulted in high caseloads (we were unable to validate applications or issue decisions for the best part of a month). However, the team was fantastic in working down the backlog—caseloads and our performance stats are now back at pre-deployment levels.”

“We are reaping the benefits of this change which has delivered significant efficiencies and improvements to our working practices. We have also embarked on the next stage of this digital journey with further enhancements, including using AI to assist in validating applications.”

This pattern is becoming familiar: short-term disruption from system change, followed by structural efficiency gains, then targeted AI to reduce friction in high‑volume, rules‑based tasks such as initial validation. Importantly, this sits alongside substantial policy work, Milton Keynes’ Proposed Submission (Regulation 19) MK City Plan 2050, supported by sustainability appraisal, consultation statements, health impact assessment, background papers and an extensive evidence base. AI does not shortcut statutory process or professional judgment. It acts as a force multiplier for the people who apply it.

[edit] Key issues across UK planning and construction and why AI matters

Capacity and delays in development management. Many LPAs continue to face high caseloads and staff shortages, which translate into validation backlogs and determination delays for major and minor applications. AI’s role is to accelerate the “front door” checks (document presence, local list conformity), highlight likely omissions, and surface similar precedent cases, allowing officers to spend time where it counts—on material considerations and negotiation.

Policy complexity and evidence overload. Plan-making now requires extensive evidence bases—sustainability appraisals, transport assessments, housing need analyses, health impact assessments and more. Officers must navigate thousands of pages across multiple studies. AI can map cross-references, extract policy-relevant insights, and show where consultations raise recurring themes by place and topic, improving traceability and reducing cognitive load.

Community trust and representation. Large consultations generate thousands of representations. Manually triaging themes, sentiment and location-specific issues is slow and can miss patterns. AI-supported clustering and topic modelling help authorities understand what communities are saying—more completely and earlier—so engagement and design can adjust before problems harden into objections.

Productivity and cost pressure in delivery. Contractors face inflation, supply volatility and thin margins, while clients demand faster, safer and greener projects. AI-enabled planning clarity up front—better risk signalling on transport, environmental constraints, design codes and conditions—reduces downstream redesign, claims and delays.

Skills shortage and institutional memory. Turnover and recruitment challenges can erode consistency. AI can help capture institutional knowledge by linking new cases to prior decisions, policy interpretations and appeal outcomes, supporting more consistent officer advice and faster onboarding.

According to a recent report by the BBC News, fewer than 29,000 projects were granted permission by councils in the year ending June 2025 - striking a blow to the government's promise to deliver 1.5 million homes by the next election.

The number of planning approvals for new homes in England is unacceptable, the new housing secretary Steve Reed, said, after official data showed permission for building homes fell to a record low during Labour's first year in office.

Only about 7,000 applications for housing were granted permission between April and June 2025 - the lowest three-month figure since records began in 1979 and an 8% fall on the same three months of 2024.

[edit] What AI is doing in planning today

• Comment synthesis at scale: clustering themes in objections and representations by geography, topic and sentiment to surface material planning considerations quickly.

• Validation assistance: checking document presence against local lists and flagging likely omissions to free officers for substantive assessment.

• Evidence navigation: mapping relationships across appraisals and technical reports to accelerate officer understanding.

• Early risk detection: spotting recurring patterns (for example daylight/sunlight or transport concerns) in time to inform design changes or targeted engagement.

[edit] Humans in the loop and way forward

Keeping humans in the loop is essential. Responsible deployments tend to be grounded in local policy and data, provide explainable outputs with source citations, and embed clear governance about where AI assists and where officers decide, with audit trails for transparency. This is about augmentation, not automation. Officers still make determinations, lead negotiations, and balance competing policy objectives.

A practical path forward

• Start with a single workflow—validation triage or comment clustering—and define success measures such as backlog days, officer hours reclaimed, error rates, appeal risks and stakeholder satisfaction.

• Use local policy and historic decisions to tune models so outputs reflect local context.

• Build verification into the process: require source citations and officer sign-off; log decisions for audit.

• Invest in people: upskill officers on data literacy, model limitations and prompt verification; celebrate internal wins to shift the narrative.

• Share lessons across authorities: publish methods and metrics so successes are replicable and pitfalls avoided.

Bottom line, this is not about automation replacing expertise. It is about augmentation—keeping humans firmly in the loop while giving them the ability to process information at scale. As Milton Keynes demonstrates, there may be bumps in the road, but the destination is a more resilient, evidence‑led and responsive planning system.

Article is written by Justin Aboh based on insights from UK Construction Week 2025's session by Architect George Clarke, and Jacqueline Glass, Dean, The Bartlett, UCL’s Faculty of Built Environment, on 'Disruption and Innovation, scaling changes in construction'.