State & Local GovernmentIssue 073 records
Citizen Engagement & Services
Citizen-engagement edition for November 4, 2026. Novosibirsk launched a university-built portal assistant that answers only from 38 official regulations. Madrid announced AI form-checking in its SOL assistant with chat handoff to the 012 line, without a start date. A Nordwest-Zeitung test found Wilhelmshaven's new city chatbot giving confident wrong answers on elections and an unexploded-ordnance site.
- Evidence records
- 3
- Cross-source patterns
- 1
- Evidence classes
- 3 independent reporting
- Outcomes
- 2 emerging1 cautionary
- Source freshness
- 3 new this fortnight
- Research completed
- 2026-11-05
Choose a role to see its takeaway beside every record in the ledger.
Patterns across the evidence
Synthesis · Lighthouse Advisory interpretation
AI front doors are being bounded by corpus and handoff design, and an independent test shows why
Novosibirsk restricts its assistant to 38 regulations with no web access; Madrid says SOL will guide but not file, with handoff to 012 agents. The NWZ test of Wilhelmshaven's site-wide bot found stale and inconsistent answers, including on safety. Together they suggest scope limits and human escalation matter as much as coverage; none publishes accuracy data.
Operating questionHave we defined what our assistant may answer, how fast it learns new content, and when it must hand off to staff?
Supporting evidenceNovosibirsk City Hall and Novosibirsk State University (reported by IT Russia)Comunidad de Madrid regional government (reported by Qué Madrid)City of Wilhelmshaven, Germany (tested by Nordwest-Zeitung)
Evidence ledger
Full record · every source keeps its link and limitations
Novosibirsk puts a university-built AI assistant on its city services portal that answers only from 38 official regulations
IT Russia reported on Nov. 4 that an AI assistant built by Novosibirsk State University on its SIGMA platform went live on the city's Novosibirsk Online portal, rebranded in late October. It helps residents find municipal services, including links to the federal Gosuslugi portal, and answers only from 38 official regulations, without internet access or outside information. The city says it is meant to reduce district administration workload. A related staff tool, SIGMA Regulations, is said by its developers to cut documentation search time by at least half. No usage or accuracy figures are reported.
Why it matters, evidence and limitations
- Why it matters
- A city-university build of a closed-corpus service assistant, a design choice U.S. cities weigh against vendor bots that index the whole website. Transferability is limited: different legal, procurement and data regimes.
- Evidence and measured results
- Trade news report quoting a city official; no measured outcomes; the staff-tool time saving is a developer claim.
- Limitations and uncertainty
- Non-US deployment; single secondary report; no accuracy, usage or cost data; the 'at least half' figure is a developer claim about a staff tool.
Comunidad de Madrid will add AI to its SOL assistant to check forms and missing documents in the Cuenta Digital, with handoff to 012 agents by chat; no start date
Qué Madrid reported on Nov. 3 that the regional government will add AI to SOL, the assistant in its Cuenta Digital account used by more than 1.5 million residents for over 350 regional services. Announced functions: answer procedure questions in real time, review entered data, detect errors and flag missing documents. SOL will guide but not file or sign on the user's behalf. The 012 citizen line will be integrated so residents can be handed to an agent by chat with context carried over. 012 channels handled more than 1.7 million queries in the first half of 2026, up 2%. No start date or rollout schedule is given.
Why it matters, evidence and limitations
- Why it matters
- Moves the AI assistant from answering questions to in-form validation inside an authenticated account, with warm handoff to the contact center. That is the design many U.S. permitting and benefits portals are considering.
- Evidence and measured results
- News report of a government announcement; contact-center volume is government data; no AI performance data.
- Limitations and uncertainty
- Announcement only: no start date, rollout schedule, vendor or cost. Non-US context. Single report.
Nordwest-Zeitung tests Wilhelmshaven's new city chatbot and finds confident wrong answers on election results, officials and an unresolved unexploded-ordnance site
A Nordwest-Zeitung reporter tested Wilhelmshaven's new AI chatbot, launched in October with a claim of over 100 languages. Asked on Oct. 31, it did not know about a public inauguration ceremony the city had announced on its own site on Oct. 30. It treated the Sept. 13 local election as a future event, then gave a wrong 'official final result' order, and in another session said results were unpublished. It misassigned council group spokespeople, varying across sessions, and said investigations of three suspected unexploded-ordnance sites were closed when one remains open. The reporter notes it also gave several helpful answers and calls the test a sample, not a full evaluation.
Why it matters, evidence and limitations
- Why it matters
- Independent follow-up on a launch the archive covered (local-government, Oct. 19): a concrete test showing stale content and confident errors, including a safety-related one. Useful evidence for acceptance testing of U.S. city chatbots.
- Evidence and measured results
- Journalist's hands-on test with specific questions and checked answers; small, non-systematic sample.
- Limitations and uncertainty
- Single reporter, handful of questions, no systematic scoring; non-US context. City response not reported.
How to read this edition
Source findings, measured results and limitations come from the cited publications. Patterns, operating questions, role takeaways and implementation considerations are Lighthouse Advisory interpretation, stated as questions to validate locally rather than guaranteed outcomes. Vendor and operator claims are labeled as claims. Full research method.
- Independent reporting
- Independent reporting with attributable sources but without a formal evaluation design.