Our Take
This is a narrow, supervised deployment: planners make the final call and audit trails are mandated, which limits both the speed gain and the liability exposure.
Why it matters
UK planners process 70% of applications for routine cases like extensions and loft conversions. Removing administrative friction on those could accelerate housing delivery without transferring decision authority to an algorithm.
Do this week
Planning authority leaders: request pilot enrollment or monitor Barnet, Camden and Dorset trial outcomes before 2027 rollout so you can assess fit for your backlogs.
DeepMind and UK Government Co-Build Planning Prototype
Google DeepMind, working with the UK government, Google Cloud, Faculty, and planning authorities in Barnet, Camden and Dorset, is building an AI-powered assistant for planning officers. The tool is designed to cut application decision times by 50% (company-stated goal). It consolidates data from planning documents, identifies relevant local and national policies, summarizes consultation feedback, and drafts the initial assessment report. The planning officer retains full authority to edit, approve or reject every decision. Decisions are logged with a complete audit trail.
The prototype builds on Extract, a tool released in June 2026 that converts legacy planning PDFs into structured data. Extract has been tested across 20+ councils and is expected to save the average council around 255 hours of manual work annually (company-reported). The planning prototype is slated for national rollout to all English councils from 2027.
Householder Applications Are the Real Bottleneck
Householder applications (extensions, loft conversions, minor alterations) account for nearly 70% of all planning applications each year but are administratively heavy: officers cross-reference policy documents, historical files and PDFs for hours on routine cases. By automating data extraction and drafting on these straightforward applications, planners free capacity to focus on complex, site-sensitive cases that require human judgment and public engagement.
The UK government's target is 1.5 million new homes by 2029. Processing delays at the local authority level are a documented constraint. Removing administrative drag on the 70% of applications that are predictable could materially accelerate the pipeline without changing policy or reducing scrutiny. The fact that officers remain the final decision-maker and the audit trail is explicit means the system is being positioned as a clerical multiplier, not a replacement for planning authority.
What Planners Should Watch
The pilot in three councils provides early evidence of whether the 50% time saving materializes and whether the tool's draft reports require heavy revision or are genuinely usable. Practitioners in other councils should review trial outcomes before 2027 rollout. Key questions: Do officers accept the tool's policy citations without verification? Does the draft assessment match actual decision patterns? Does the audit trail satisfy both internal quality control and public FOI requests?
The deployment model (planner retains authority, decisions are audited) is conservative and administratively sound but also limits the speed gain. If 80% of a decision's time is spent on drafting and 20% on final review, cutting 50% of total time requires the tool to handle most of the drafting and some review overhead. That implies trust in the model's output, which the audit requirement suggests is still being earned.