evidoria

← Back to browse

Good practice Imported

Extract — Google DeepMind's Gemini Tool for Digitising the UK's Paper Planning Archives

United Kingdom · London · See the United Kingdom profile · See the London profile

Evidence: Observational / pre–post Top 6% 87/100 · Ask Evidence Copilot about this practice

A Gemini-powered tool converts decades of paper planning records into structured digital data, saving councils an estimated 255 hours a year each; after piloting with 20+ local authorities it went live for every council in England in June 2026.

255 hours per council per year
Estimated staff time saved
20 authorities (20+)
Local planning authorities piloted
Extract — Google DeepMind's Gemini Tool for Digitising the UK's Paper Planning Archives

Details

Promoter
i.AI (UK Government AI Incubator, Department for Science, Innovation and Technology) with Google DeepMind
Period
2025–2026 (national rollout June 2026)
Keywords
AI incubator, planning & housing, digital government, document digitisation

Context

Extract is a tool built jointly by the UK government's AI incubator (i.AI, part of the Department for Science, Innovation and Technology) and Google DeepMind to convert England's backlog of paper-based planning records, including scanned PDFs, handwritten notes and old maps, into structured digital data.

Activities

The tool uses Gemini's multimodal reasoning to read documents, extract fields, trace property boundaries as polygons, and georeference historical maps against current mapping, work that previously required a planning officer to manually re-key each document by hand.

Results

After trials across more than 20 local planning authorities, Google DeepMind's June 2026 account estimated savings of around 255 hours of manual work per council per year and described the tool converting hundreds of pages of legacy documents into usable data in minutes, and Extract was made available to every council in England that same month as part of a push to support a target of 1.5 million new homes.

Conclusions

Independent coverage frames Extract as one tool within a wider planning-AI programme rather than a finished product, since neither the official announcement nor independent reporting has yet published error rates, extraction-accuracy figures, or an independent audit of the tool's outputs.

Implementation
Indicative cost
Medium (€50k–€500k)
Time to results
Short (< 1 year)
Staffing & skills
i.AI (UK Government AI Incubator, Department for Science, Innovation and Technology), Google DeepMind engineering team

Conditions for success

  • Piloting across more than 20 local planning authorities with varied IT environments and document archives before national rollout
  • A joint public-private delivery model pairing a named government AI incubator with a named private AI lab
  • An explicit national housing-policy goal (1.5 million new homes) providing political backing for scale-up

Common failure modes

  • No published error-rate or extraction-accuracy figures, so the case for impact currently rests on time-saved estimates rather than validated data quality

Where it fits

Governance type
national
Scale
national
Income level
high-income

Do you run this practice? Claim it — verified implementers get a public contact pathway and can propose corrections.

Data sources

Where this practice's information was retrieved from, and when.

Attachments

Similar practices you may find useful