{"resourceId":"nsw-hunter-councils-daisy-ai-planning-assistant-16533-conversations-wingecarribee-removal-20261030","versions":[{"version":"external-a32da5a92b231233a463e024be427b23e7f9fdafeeffd3a36fbe8983fe1c9179","resource":{"id":"nsw-hunter-councils-daisy-ai-planning-assistant-16533-conversations-wingecarribee-removal-20261030","title":"Newcastle and Muswellbrook councils launch the DAISY AI planning assistant; vendor reports 16,533 conversations in nine months while Wingecarribee removes it after a pilot","organization":"City of Newcastle and Muswellbrook Shire Council (New South Wales); vendor Enterprise AI","sector":"Municipal planning and development applications","geography":"Hunter region, New South Wales, Australia","publishedAt":"30 October 2026 (Information & Data Manager)","publicationDate":"2026-10-30","eventDate":null,"sourceName":"Information & Data Manager (IDM)","sourceLabel":"Trade press reporting drawing on vendor data","sourceUrl":"https://idm.net.au/article/0015850-hunter-councils-add-ai-planning-enquiries","evidenceClass":"vendor-claim","outcomeClass":"mixed","topics":["knowledge-work","operating-model","governance-procurement"],"finding":"Two more New South Wales councils, Newcastle and Muswellbrook, put the DAISY AI planning assistant on their websites to explain which planning rules apply to a property and which documents an application needs. Across all live councils the vendor reports 16,533 conversations from January to September 2026, a third of them outside counter hours, and models 4,133 staff hours returned at 15 minutes per enquiry. It also claims a 20% cut in assessment timeframes and a 49% rise in service capacity for early adopters, without a stated method. The same article notes Wingecarribee Shire Council is removing the tool after a 15-month pilot.","sledRelevance":"Why now: published 30 October with new council launches and nine-month usage totals. It is a rare case where adoption and withdrawal of the same municipal AI tool appear side by side, which is useful for U.S. cities weighing permit-guidance assistants. Transferability is limited by NSW planning law and vendor-only metrics.","evidence":"Vendor-reported conversation counts and modelled staff hours; unattributed assessment-time and capacity claims; one sentence on a council removing the tool. No independent evaluation.","architectureImplications":"Resident-facing guidance is the first stage; the vendor is integrating with Civica Altitude and Muswellbrook is exploring planner-side assessment support, which would move the tool from public information into the case-management system.","governanceImplications":"The vendor states planners remain responsible for all assessments and determinations. Councils moving to assessment support would need rules on how AI-prepared compliance checks and report drafts are reviewed.","securityPrivacyImplications":"The article does not describe what resident data conversations capture or how it is retained.","caveats":"Vendor metrics; hours modelled; 20% and 49% claims unattributed; withdrawal reason not reported.","streamIds":["local-government"],"roles":{"sales":"Framing: an AI planning assistant is spreading across NSW councils while one council is removing it after a pilot. Interpretation: planning departments with heavy pre-lodgement enquiry loads may welcome a conversation about applicant guidance, but the Wingecarribee exit means any pitch should start from agreed success measures. The evidence is vendor counts and modelled hours, so it supports a scoped pilot, not promises of faster assessments or saved staff time.","engineering":"Framing: DAISY starts as website guidance and is moving toward planner-side checks through a Civica Altitude integration. Interpretation: engineers should confirm which planning controls and property data the assistant draws on, how rule updates are loaded, and what happens when it lacks an answer. A proof of value should test answer accuracy on sampled real enquiries, because the source reports volume and modelled hours but no accuracy results.","delivery":"Framing: councils are adopting the same tool in stages, and at least one ended it after 15 months. Interpretation: delivery teams should set exit criteria and a review point at the start, track enquiry deflection against actual phone and counter volumes, and decide who maintains the rule content. Acceptance criteria here are proposals; the source does not report measured outcomes for any named council."},"roleDetails":{"sales":{"customerProblem":"Planning teams spend time on avoidable enquiries and incomplete development applications.","stakeholders":["Planning director","Customer service manager","Council IT lead","Elected councillors"],"discoveryQuestions":["How many pre-lodgement enquiries arrive by phone and counter each month?","What share of applications are returned as incomplete?","What would make the council end a pilot early?","Who maintains planning-rule content today?"],"valueHypothesis":"Bounded: after-hours guidance could reduce some routine enquiries if answers prove accurate.","potentialEngagement":"A pilot design with baseline enquiry counts, accuracy sampling and exit criteria for an AI planning-guidance assistant.","unsupportedClaims":["Assessment times fell 20% at any named council","Staff hours were actually saved","Answers are accurate"]},"engineering":{"fit":"Councils with codified planning controls and public property data.","architectureIntegration":"Website assistant grounded in planning controls, with a planned link to the case-management platform for planner-side checks.","prerequisites":["Machine-readable planning controls","Property data access","Content update process"],"constraints":["NSW-specific planning law","Vendor-reported metrics only","Integration tied to one platform vendor"],"security":"Define what conversation data is stored and for how long before launch.","proofOfValue":["Score answers on a sample of real enquiries against planner judgement","Compare incomplete-application rates before and after launch"]},"delivery":{"work":"Load planning content, launch applicant guidance, monitor usage and decide on planner-side extension.","dependencies":["Vendor content updates","Planner time for answer review","Platform integration if extended"],"ownership":"Council planning department, with customer service for resident-facing performance.","skillsAdoption":"Planners learn to review and correct the assistant's rule content.","governanceCheckpoints":["Pilot review against agreed measures","Decision gate before planner-side use","Annual contract and exit review"],"acceptanceCriteria":["Proposed: sampled answer accuracy above an agreed threshold","Proposed: measured drop in phone and counter enquiries","Proposed: documented exit plan"],"risks":["Outdated rule content gives wrong guidance","Metrics remain vendor-only","Tool removed after sunk effort"]}},"retrievedAt":"2026-10-31T22:18:25-05:00","enrichedAt":"2026-10-31T22:22:10-05:00","enrichmentBasis":"retrieved source","accessibilityWorkforceImplications":"A third of conversations happened outside counter hours, pointing to after-hours demand; staff time effects are modelled, not observed.","procurementImplications":"One council ended the tool after a pilot, so contracts should include exit terms and agreed success measures before rollout.","operatingModelImplications":"Councils are staging from applicant guidance to planner assistance, which changes who owns the tool from communications to planning.","sourceVerification":{"openedUrl":"https://idm.net.au/article/0015850-hunter-councils-add-ai-planning-enquiries","referenceExcerpt":"Enterprise AI estimates the conversations returned 4,133 hours of staff time.","promptVersion":"sled-research-v4.0","model":null,"basis":"agent-reported inspection"}}}]}