{"resourceId":"norfolk-e911-aws-ai-nonemergency-lines-20261005","versions":[{"version":"external-2211ee8029a3c39eefd8b0a7aa687d7ac7a7c34ea7cfd323e17c3a2f08ee3ae9","resource":{"id":"norfolk-e911-aws-ai-nonemergency-lines-20261005","title":"Norfolk E911 Division puts AWS-powered AI on non-emergency lines, projecting 35–40% call diversion while 911 stays human","organization":"Norfolk Police Department Emergency 911 (E911) Division, City of Norfolk, Virginia","sector":"Municipal PSAP / emergency communications (police, fire and medical calls)","geography":"United States (Virginia)","publishedAt":"2026-10-05","publicationDate":"2026-10-05","eventDate":"2026-10-05","sourceName":"City of Norfolk News Flash","sourceLabel":"Official city announcement of the E911 Division's AI non-emergency line launch","sourceUrl":"https://www.norfolk.gov/m/newsflash/home/detail/10535","evidenceClass":"government-evaluation","outcomeClass":"emerging","topics":["operating-model","accessibility-workforce","infrastructure"],"finding":"The City of Norfolk announced on Oct 5, 2026 that its Police Department's E911 Division launched an Amazon Web Services AI-powered service for select non-emergency and informational calls. E911 staff answer over 250,000 police, fire and medical calls and about 150,000 administrative calls a year. The city expects the system to divert roughly 35–40% of non-emergency calls immediately. Callers who say 'I want to speak to an agent' are redirected to a non-emergency line for the next available call taker; all 911 calls stay with trained telecommunicators.","sledRelevance":"A large municipal PSAP that answers fire and medical as well as police calls is adopting a cloud contact-center AI front door; included here for its 911-center operating model (cross-stream with public safety because the division sits in the police department).","evidence":"Official agency announcement, not an evaluation (closest schema class used). The 35–40% diversion figure is a forecast; no baseline hold times, accuracy, misrouting or post-launch data are published.","architectureImplications":"Cloud contact-center AI (AWS) sits on the non-emergency trunk, answers FAQs, and hands off to the human non-emergency queue on request; 911 trunks are untouched.","governanceImplications":"Launch without published success metrics or review cadence; the agency should define how diverted, abandoned and misdirected emergency calls on the non-emergency line are counted before claiming the forecast.","securityPrivacyImplications":"Announcement does not describe recording, transcript retention, or AWS data-use terms; these need to be set out for a public-safety call center.","caveats":"Forecast only; launch-day announcement; the 'speak to an agent' path returns callers to the human queue rather than an immediate answer; no detail on emergency detection on non-emergency lines.","streamIds":["emergency-services"],"roles":{"sales":"Frame Norfolk as a capacity story with numbers attached: about 150,000 administrative calls a year on top of more than 250,000 police, fire and medical calls, and a forecast that AI absorbs 35–40% of non-emergency calls. The useful conversation is how the center will prove the forecast and what happens to callers who ask for a human. Offer measurement and configuration help, not guaranteed hold-time savings, since nothing has been measured yet.","engineering":"This is a general cloud contact-center deployment on non-emergency lines, with a spoken opt-out that drops callers into the existing human queue. Engineers should confirm how the AI recognizes an emergency spoken on a non-emergency line, how transfers to 911 work, what call data AWS retains, and how intents map to city services. None of that is in the announcement, so test with scripted calls before scaling.","delivery":"Run this as a measured rollout: capture pre-launch hold and abandonment baselines, log every opt-out and emergency transfer, and review a sample of AI-handled calls weekly. Publish a 90-day result against the 35–40% forecast. Assign one owner for FAQ content so answers stay current as services change, and keep a written rule for transferring any caller who describes an emergency, since the release does not say how that works."},"roleDetails":{"sales":{"customerProblem":"Administrative calls wait behind 911 traffic, producing long non-emergency hold times.","stakeholders":["E911 manager","Police and fire leadership","City IT","City council","Residents needing human access"],"discoveryQuestions":["What are current non-emergency hold and abandonment times?","How will diversion be measured against the forecast?","How are emergencies on non-emergency lines detected?","What data does the cloud provider keep?"],"valueHypothesis":"An AI front door may answer a meaningful share of routine questions immediately, if diversion and safe escalation are measured.","potentialEngagement":"90-day measurement plan and call-sample review for an AI non-emergency line.","unsupportedClaims":["35–40% diversion achieved (forecast only)","Faster 911 answer times","Lower staffing need"]},"engineering":{"fit":"FAQ and routing agent on non-emergency PSAP lines.","architectureIntegration":"Non-emergency trunk to AWS AI service; spoken opt-out to human non-emergency queue; 911 trunks unchanged.","prerequisites":["Intent and FAQ catalog","Emergency-transfer rule","Call logging and baselines"],"constraints":["No published accuracy data","Opt-out returns caller to queue","Voice-only interaction"],"security":"Define transcript retention, encryption and provider data-use limits in the contract.","proofOfValue":["Measured diversion vs 35–40% forecast","Emergency transfer accuracy on test calls","Non-emergency hold time before and after"]},"delivery":{"work":"Baseline, launch, log opt-outs and transfers, review samples, publish results.","dependencies":["AWS configuration","Telephony","City service content owners"],"ownership":"Norfolk Police E911 Division operations.","skillsAdoption":"Telecommunicators review AI call samples and flag wrong answers or missed emergencies.","governanceCheckpoints":["Pre-launch emergency transfer test","30-day sample review","90-day public results"],"acceptanceCriteria":["Every emergency spoken on test calls transferred","Diversion measured against forecast","Opt-out honored on all requests"],"risks":["Missed emergency on non-emergency line","Outdated FAQ answers","Callers frustrated by return to queue"]}},"retrievedAt":"2026-10-05T22:06:48-05:00","enrichedAt":"2026-10-05T22:06:48-05:00","enrichmentBasis":"retrieved source","accessibilityWorkforceImplications":"Opt-out to a human is offered by voice command; callers who cannot use voice prompts or who speak other languages are not addressed. Framed as freeing telecommunicators for critical incidents, not as staff reduction.","procurementImplications":"Uses a hyperscaler contact-center stack rather than a 911-specialist vendor, so contract terms on data, uptime and configuration support should be checked against PSAP requirements.","operatingModelImplications":"AI front door for administrative volume, human queue as fallback, 911 fully human.","sourceVerification":{"responseId":"curl-norfolk-e911-2026-10-05T22:06:48-05:00","openedUrl":"https://www.norfolk.gov/m/newsflash/home/detail/10535","referenceExcerpt":"The Amazon AWS system is expected to divert roughly 35-40 percent of the non-emergency-related calls immediately upon going into use.","promptVersion":"sled-research-v4.0","model":null,"basis":"agent-reported inspection"}}}]}