State & Local GovernmentIssue 094 records
Citizen Engagement & Services
Citizen-engagement edition for November 6, 2026. Orland Park paused its Humphrey AI phone agent after transfer failures, publishing September metrics (2,687 calls; 71% transferred; 6% negative) while remediating department numbers and funding Gemini staff archive support. Chattanooga's mayor reported an AI civic-dialogue pilot with Change.org and Jigsaw that drew nearly 2,000 submissions and lifted participant optimism from about 60% to 82.6%, and cited Nerdery agentic AI for residential and electrical permits. Innopolis is testing multilingual Inna for Telegram, web, voice and phone with human concierge takeover toward an end-of-November launch. A DeepMind case study describes Ukraine's on-prem Gemma ePermit recommendations for veterinary licenses with officials retaining decisions.
- Evidence records
- 4
- Cross-source patterns
- 1
- Evidence classes
- 3 independent reporting1 vendor claim
- Outcomes
- 2 emerging1 cautionary1 mixed
- Source freshness
- 3 new this fortnight1 undated
- Research completed
- 2026-11-07
Choose a role to see its takeaway beside every record in the ledger.
Patterns across the evidence
Synthesis · Lighthouse Advisory interpretation
Resident-facing AI is shipping with human gates—and operators are publishing service metrics when channels break or engage
Orland Park paused Humphrey after transfer failures and published September call volume and sentiment while fixing DIDs before resume. Chattanooga reported participant optimism rising from about 60% to 82.6% in an AI civic dialogue and described Nerdery agentic permitting with human-in-the-loop governance. Innopolis's Inna keeps concierge takeover during testing toward an end-of-November launch. Ukraine's ePermit Gemma path recommends issue/reject/manual review to officials under an on-prem constraint. Together they show human final authority plus early operational metrics—not yet independent outcome evaluations.
Operating questionWhere does our resident AI stop, who can residents always reach, and which service metrics must be public before we resume or expand?
Supporting evidenceVillage of Orland Park, IllinoisCity of Chattanooga (with Change.org and Google Jigsaw; Nerdery for permitting)City of Innopolis, Republic of TatarstanMinistry of Economy, Ecology and Agriculture of Ukraine (Google DeepMind case study)
Evidence ledger
Full record · every source keeps its link and limitations
Orland Park pauses Humphrey AI phone agent after transfer failures; September metrics show 2,687 calls, 71% transferred, 6% negative, while board funds Gemini staff archive support
On Nov. 6, 2026 GovTech republished Daily Southtown reporting that Orland Park paused its Humphrey AI phone agent after October resident complaints that callers could not reach a person. IT Director Tad Spencer said pre-launch tests missed the failure: the system could not transfer callers to the operator or Parks and Recreation because of associated phone numbers, and the village is working with a vendor to add numbers before resume. Of 2,687 September calls, 71% transferred to a department; 91% were perceived neutral and 6% negative. The board also approved a $43,200 Zion Cloud Solutions contract for Google Gemini staff data-archive support (40 hours/month from January), with Spencer citing 97% of staff questions answered and ~$85,000 prior archive cost avoided. Mayor Dodge said the fault was in the phone system, not the AI technology.
Why it matters, evidence and limitations
- Why it matters
- A Chicago-suburb 311-style AI phone front door that already appeared in the archive as a resident transfer failure now has measured call volume, sentiment, an explicit pause, and a vendor fix plan—exactly the human-transfer control CE tracks for resident voice agents.
- Evidence and measured results
- Syndicated local reporting with named IT director, September operational metrics, pause confirmation, and board contract details; village self-reported sentiment labels.
- Limitations and uncertainty
- Sentiment percentages are village-perceived labels; no independent accuracy audit of Humphrey answers; Gemini savings are staff-side, not resident phone outcomes.
Chattanooga reports AI civic-dialogue pilot drew ~2,000 submissions and lifted participant optimism from ~60% to 82.6%; mayor also cites Nerdery agentic AI for residential and electrical permits
On Nov. 6, 2026 PublicTechnology reported Mayor Tim Kelly's account of Chattanooga's AI-driven civic dialogue with Change.org and Google Jigsaw on parks and transportation priorities. The mayor said the interactive AI dialogue yielded nearly 2,000 qualitative submissions totaling roughly 350,000 words, with follow-up virtual civic forums validating input, and that participant optimism that 'residents can talk through problems together' rose from around 60% before to 82.6% after. He framed the method as an alternative to policy driven by a few loud council voices. Separately he said the city is working with Nerdery on an agentic AI tool to assess and issue residential modification and electrical permits by reviewing drawings and plans, alongside internal Google Gemini tools under a human-in-the-loop governance model. Measured permitting cycle-time results were not published.
Why it matters, evidence and limitations
- Why it matters
- Same-day CE signal combining measured resident-engagement trust lift with an explicit agentic permitting-review build—covering constituent digital engagement and permitting AI in one primary interview write-up.
- Evidence and measured results
- UK trade interview of the mayor citing city-reported survey volumes and optimism percentages; city July 22 pilot announcement confirms partners; no independent evaluation or permitting metrics.
- Limitations and uncertainty
- Optimism lift is among participants, not a citywide random sample; permitting AI is described as in progress without measured speed or error rates; Liverpool summit remarks are self-reported.
Innopolis unveils Inna, a multilingual AI resident agent for Telegram, website, voice and phone, in testing with expected launch by end of November 2026 and human concierge takeover
On Oct. 30, 2026 the City of Innopolis said Mayor Ildar Khuzzyatov unveiled Inna, an AI virtual assistant for residents via messaging apps such as Telegram and the official website, also handling voice commands and telephone calls in Russian, Tatar, and English. The system uses an LLM over a knowledge base from the city website and official social channels, updated automatically; the concierge team trains Inna and can join the chat when the agent cannot resolve an issue. The city claims personalized history, follow-up questions, routine-inquiry handling with wait times under one minute, and analysis of requests to spot negative trends. Development is in testing with expected launch by the end of November 2026. A Telegram chatbot named Inna has operated since 2017 for amenities and service reports; the digital twin monitoring ~100,000 assets is described separately.
Why it matters, evidence and limitations
- Why it matters
- A primary city release on a multilingual, multi-channel resident AI agent with explicit human concierge takeover and a near-term public launch window—core CE virtual-agent scope, not yet in the CE archive.
- Evidence and measured results
- Official city English press release with named mayor; aspirational wait-time and trend-analysis claims; no measured accuracy or diversion metrics yet.
- Limitations and uncertainty
- Self-reported city announcement; 'first in Russia' claim unverified; under-one-minute wait and trend detection are design goals during testing, not published results.
Ukraine Ministry integrates on-prem Gemma 3 27B into ePermit licensing review for veterinary licenses, recommending issue, reject, or manual review while officials retain decisions
A Google DeepMind Gemmaverse case study describes Ukraine's ePermit service on the Diia portal integrating Gemma 3 27B for license application review after Qwen 2.5 7B VL extracts text from scanned documents and registry data are combined. Legal constraints require closed infrastructure with no external API calls or cloud dependencies; the model runs on-premise via vLLM on NVIDIA H100 GPUs in a Ukrainian datacenter. Gemma recommends one of three outcomes to the assigned official—issue the license, reject, or further manual review—for veterinary licenses initially, with expansion to other types planned. Officials quoted say Gemma balanced reasoning and on-prem deployment; the write-up claims dramatically cut processing time, no latency issues, and fewer rejections/resubmissions/workload, without publishing numeric baselines. Decisions are described as auditable via logged inputs, outputs, and rules.
Why it matters, evidence and limitations
- Why it matters
- Licensing/permitting AI with an explicit on-prem, no-external-API constraint and human final decision—transferable operating constraint for SLED agencies that cannot send applicant documents to public clouds.
- Evidence and measured results
- Vendor case study with named ministry CDTO/AI advisor quotes; impact claims lack numeric baselines or independent evaluation; retrieval date used as publishedAt because the page is undated.
- Limitations and uncertainty
- DeepMind marketing page; 'dramatically' faster processing without numbers; veterinary-only production scope; English typos in ministry name on the page.
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.
- Vendor claim
- A supplier-provided assertion that has not been upgraded to independent evidence.