{"resourceId":"mhclg-planai-consultation-preparation-2026","versions":[{"version":"external-22d54ed05e636be4bb0be67f1d24f31ae196c27b747fa9efa6b1135d1a25bed3","resource":{"id":"mhclg-planai-consultation-preparation-2026","title":"PlanAI pilot separates rapid summarisation from weeks of preparation and quality assurance","organization":"Ministry of Housing, Communities and Local Government; participating English planning authorities","sector":"Local-plan consultation analysis","geography":"England; transferable workflow questions, not U.S. planning rules or savings estimates","publishedAt":"July 30, 2026","publicationDate":"2026-07-30","eventDate":null,"sourceName":"MHCLG Digital","sourceLabel":"Government programme team's retrospective pilot account","sourceUrl":"https://mhclgdigital.blog.gov.uk/2026/07/30/using-ai-to-support-faster-local-plan-consultation-analysis/","evidenceClass":"government-evaluation","outcomeClass":"mixed","topics":["knowledge-work","data-security","accessibility-workforce","operating-model"],"finding":"The official pilot account reports large analysis-stage efficiencies alongside preparation costs and inaccurate policy tagging.","sledRelevance":"Interpretation: Relevant to municipal consultation teams processing resident comments; summarisation must preserve participation and minority concerns.","evidence":"Five authorities extended the pilot. MHCLG reports around 90% efficiency gains but says preparation and iteration took two to three weeks for some authorities. Greenwich encountered inaccurate policy tags. The account does not provide comparable denominators, total workflow costs or an independent quality benchmark.","architectureImplications":"Interpretation: Separate input preparation, tagging, summarisation and reviewer approval in both the pipeline and performance logs.","governanceImplications":"Interpretation: Require a trace from thematic conclusions to resident submissions, with a route to challenge omitted or misrepresented views.","securityPrivacyImplications":"Interpretation: Redact personal data before processing and verify whether deleted source text persists in prompts, caches or logs.","caveats":"Programme-authored narrative rather than a controlled study. The reported efficiency applies to a component. No demonstrated end-to-end causal benefit or accessibility evaluation.","streamIds":["local-government"],"roles":{"sales":"Interpretation: Engage the consultation lead, information governance team and community-engagement staff around whether analysis capacity is delaying plan development. Ask how submissions arrive, how minority views are represented and what a corrected thematic report costs to produce. Offer a bounded comparison on a completed consultation, with the existing manual analysis retained as a reference. A credible hypothesis is less repetitive grouping while preserving accurate representation; it is not a promise of a 90% shorter consultation. Establish whether the buyer can fund preparation and review. Clarify that an assistant's thematic output does not replace the municipality's policy judgement or engagement obligations.","engineering":"Interpretation: Build a test corpus spanning short, long, multilingual and overlapping-policy responses. Keep original identifiers through redaction, segmentation, tagging and summary generation. Prerequisites include an agreed policy taxonomy, lawful data access and reviewer-labelled examples. Evaluate omitted concerns, incorrect policy assignments and distortion of response volume, not just processing speed. Include adversarial instructions embedded in comments as untrusted data. Proposed validation should compare error rates and total analyst effort with the manual baseline, then repeat after configuration changes. Hosting and agent capabilities were not evaluated here; avoid autonomous updates to official consultation conclusions until the authority has validated a controlled approval boundary.","delivery":"Interpretation: The consultation service manager should own completeness and fairness, supported by records staff, analysts and privacy reviewers. First establish a manual reference sample and an accessible alternative route; then rehearse redaction, correction and publication. Dependencies include policy clarity and sufficient review capacity. Train staff to inspect underlying responses when a summary is ambiguous. Proposed acceptance requires every sampled theme to be traceable, all high-impact omissions to be resolved, and the manual route to receive equivalent review. Track elapsed time from receipt through approved analysis. These are proposed checks. Risks include undercounting preparation, amplifying frequent views and unintentionally excluding residents who avoid AI."},"retrievedAt":"2026-09-09T03:00:44Z","enrichedAt":"2026-09-09T03:04:04Z","enrichmentBasis":"retrieved source","accessibilityWorkforceImplications":"The account describes an AI opt-out with manual analysis. Interpretation: test equal treatment across both routes.","procurementImplications":"Interpretation: Include preparation, human checking and support costs in quotations; require exportable response-to-summary links.","operatingModelImplications":"Interpretation: Fund consultation analysts' verification work and escalation, not merely the model runtime.","updateExplanation":"New-to-archive July source; newly relevant as direct evidence of preparation and tagging limits in municipal knowledge work. No newer release or outcome claimed.","sourceVerification":{"openedUrl":"https://mhclgdigital.blog.gov.uk/2026/07/30/using-ai-to-support-faster-local-plan-consultation-analysis/","referenceExcerpt":"There are efficiency gains from using the tool but quality assurance needs to be planned as part of the workflow.","promptVersion":"sled-research-v3.1","model":null,"basis":"agent-reported inspection"}}}]}