{"resourceId":"akron-documenters-ava-nonemergency-ai-2026","versions":[{"version":"external-6b34f2abc836bf4ab615519e1bc2804f4d76681af817e09a7439adaea78d3206","resource":{"id":"akron-documenters-ava-nonemergency-ai-2026","title":"Akron dispatchers tell residents Ava AI handles non-emergency calls; ~39k fully automated","organization":"Akron Combined Communications Center","sector":"911 / PSAP emergency communications (combined fire and police dispatch)","geography":"Akron, Ohio, United States","publishedAt":"2026-09-09","publicationDate":"2026-09-09","eventDate":"2026-08-26","sourceName":"Signal Akron (Akron Documenters)","sourceLabel":"Independent local Documenters meeting record (paraphrased notes with quoted supervisor remarks)","sourceUrl":"https://signalakron.org/residents-raise-questions-about-artificial-intelligence-usage-at-akron-emergency-dispatch-center/","evidenceClass":"independent-reporting","outcomeClass":"mixed","topics":["operating-model","accessibility-workforce","governance-procurement","infrastructure"],"finding":"Signal Akron Documenters (published Sept 9, 2026) record an Aug 26 Ward 7 meeting where Akron dispatch said Ava AI is used for non-emergency—not 911—routing since May 2025 (still trial), citing 84,149 AI-handled calls with nearly 39,000 fully automated and 1,200/17,322 (7%) needing callbacks, while residents questioned wait times and whether AI pushed people to dial 911.","sledRelevance":"Why now: Sept 9 local measured non-emergency AI disclosure after the 2026-09-16 La Vergne AVA deployment edition—adds public-trust scrutiny plus callback-rate figures absent from many launch stories.","evidence":"Documenters notes: 62 workers; 330,000 calls/year; Ava since May 2025 trial; not on 911; 84,149 AI-handled; ~39,000 fully automated; 17,322 completed by AI agents with 1,200 (7%) callbacks (period unclear); frustration/cursing escalation to agent; residents asked about 911 spillover; presenters acknowledge dislike of AI.","architectureImplications":"Non-emergency conversational agent (Ava) for routing/info collection with escalation on frustration; separate 311 weekend coverage; independent of Summit County’s system.","governanceImplications":"Publish that 911 is human-only, clarify metric time windows, and study whether non-emergency AI increases 911 dialing—as residents asked.","securityPrivacyImplications":"AI collects name, phone, address, and call reason—apply retention and vendor training limits; document escalation when callers refuse AI.","caveats":"Documenters paraphrased meeting notes (not investigative audit); callback metric period unclear; combined center includes policing workloads—relevance for stream is fire/EMS-serving dispatch capacity and non-emergency AI controls.","streamIds":["emergency-services"],"roles":{"sales":"Frame Ava as a still-in-trial non-emergency router under public scrutiny—not a finished staffing fix. Discovery should pin down the callback metric window, 911 spillover monitoring, and how fire/EMS non-emergency demand is affected. Offer a community-facing metrics brief and QA sampling engagement rather than an expansion pitch. Unsupported claims include proven faster 911 answer times, safe elimination of non-emergency staff, or generalizable 7% callback performance without a defined denominator period.","engineering":"Fit is non-emergency voice AI with escalation on frustration/cursing and optional full automation for some call types. Architecture must hard-separate 911 trunks, log AI vs human handling, and support callback workflows. Prerequisites include clear intent taxonomy and public scripts. Constraints: trial status since May 2025 and resident distrust. Proof of value is audited automation share, callback rate with dates, 911 spillover rate, and telecommunicator presence on emergency lines—not raw 84k counts alone.","delivery":"Keep 911 human-only, publish scope, sample automated sessions, and answer resident spillover questions with data. Ownership stays with the combined communications center supervisor; PIO handles public explanation. SkillsAdoption trains staff to reclaim frustrated callers. Governance checkpoints include trial exit criteria and quarterly public metrics. Risks include treating paraphrased meeting stats as audited KPIs and ignoring 911 wait-time complaints blamed on AI."},"roleDetails":{"sales":{"customerProblem":"Combined PSAPs adopt non-emergency AI under staffing pressure while residents distrust automation and lack published spillover metrics.","stakeholders":["Dispatch supervisors","Fire/EMS leadership","City council/ward offices","Residents/311 users"],"discoveryQuestions":["What exact dates cover the 84k/39k/7% figures?","Is 911 spillover measured?","What are trial exit criteria?","How are fire non-emergency calls classified?"],"valueHypothesis":"If automation reduces non-emergency load without increasing 911 misuse or unsafe callbacks, trial continuation may be justified—without claiming emergency performance gains.","potentialEngagement":"Public metrics brief plus bounded QA sampling of automated sessions.","unsupportedClaims":["Faster 911 answers proven","7% callback as standing SLA","Resident acceptance","Safe staff cuts"]},"engineering":{"fit":"Non-emergency Ava agent for routing/info with human escalation; 911 remains human.","architectureIntegration":"Non-emergency trunks → Ava → CAD/route or agent; frustration detection to human; separate 311 weekend path.","prerequisites":["911 trunk separation","Callback workflow","Logging of AI-handled vs automated-complete","Public script"],"constraints":["Trial since May 2025","Metric period ambiguity","Combined police/fire center politics"],"security":"Protect collected identity/address; limit vendor training; authenticate agent handoff.","proofOfValue":["Dated automation and callback rates","911 spillover rate","Emergency-line occupancy","Resident complaint themes"]},"delivery":{"work":"Maintain human 911, publish Ava scope, sample automated calls, report spillover metrics to council.","dependencies":["QA capacity","CAD/reporting extracts","311 weekend staffing policy"],"ownership":"Communications center supervisor owns trial controls; council liaison owns public questions.","skillsAdoption":"Train reclaim of frustrated callers and explanation of when to use 311 vs non-emergency vs 911.","governanceCheckpoints":["Trial exit review","Quarterly public metrics","Spillover investigation if 911 wait complaints rise"],"acceptanceCriteria":["Dated KPI definitions","No AI on 911 trunks verified","Published callback and spillover figures"],"risks":["Meeting-note stats treated as audit","Resident bypass to 911","Unclear metric windows"]}},"retrievedAt":"2026-09-18T03:16:05Z","enrichedAt":"2026-09-18T03:20:28Z","enrichmentBasis":"retrieved source","accessibilityWorkforceImplications":"Residents dislike talking to AI; frustration escalation exists but accessibility for disabled/LEP callers is not evaluated in the notes.","procurementImplications":"Trial since May 2025 without published exit criteria in the notes—procurement should demand dated KPIs and spillover monitoring before renewal.","operatingModelImplications":"Non-emergency AI plus weekend 311 automation changes how the public reaches humans; emergency model remains human call-taking per officials.","sourceVerification":{"openedUrl":"https://signalakron.org/residents-raise-questions-about-artificial-intelligence-usage-at-akron-emergency-dispatch-center/","referenceExcerpt":"Nearly 39,000 of those calls were handled completely automated.","promptVersion":"sled-research-v4.0","model":null,"basis":"agent-reported inspection"}}}]}