{"resourceId":"jonesboro-lezle-ai-nonemergency-calltaker-2026","versions":[{"version":"external-ed4523b728198ac444f11d27b38d6885269f60ee708da28c401bfbe46ad0b889","resource":{"id":"jonesboro-lezle-ai-nonemergency-calltaker-2026","title":"Jonesboro E-911 trials Lezle AI for non-emergency calls with human monitoring and 911 ban","organization":"Jonesboro E-911 Dispatch Center","sector":"911 / PSAP emergency communications (serving police and fire agencies)","geography":"Jonesboro, Arkansas, United States","publishedAt":"2026-09-02","publicationDate":"2026-09-02","eventDate":"2026-09-02","sourceName":"KAIT","sourceLabel":"Independent local television reporting","sourceUrl":"https://www.kait8.com/2026/09/02/jonesboro-dispatch-center-is-testing-an-ai-call-taker/","evidenceClass":"independent-reporting","outcomeClass":"emerging","topics":["operating-model","accessibility-workforce","governance-procurement","knowledge-work"],"finding":"On 2026-09-02 KAIT reported Jonesboro's two-day trial of Lezle, an AI non-emergency call taker monitored by dispatchers: the center cites thin staffing (as few as three on shift; 22 full-time supporting multiple police and 11 fire departments), shows administrative call handling with follow-up SMS, and states the AI will never answer a 911 call and will escalate true emergencies to a human.","sledRelevance":"Why now: newly relevant after 2026-09-14 Weber coverage of non-emergency AI call share and trust; Jonesboro supplies a same-period U.S. PSAP control pattern—explicit 911 exclusion plus live human monitoring—useful for fire/EMS-serving dispatch centers evaluating similar tools.","evidence":"Article documents Lezle naming, staffing figures, fingerprinting example with SMS fee/hours, human monitoring during trial, Hundley quotes on never answering 911 and emergency handoff, and planned overnight continuation. No accuracy metrics.","architectureImplications":"Non-emergency agent with barge-in monitoring, emergency intent escalation to human queues, SMS side-channel for administrative instructions, and CDR/CAD notes distinguishing AI-handled calls.","governanceImplications":"Codify never-answer-911 policy, require monitored trial periods before overnight unmanned AI, and communicate scope to served fire and police agencies.","securityPrivacyImplications":"Capture and retain AI dialogs under PSAP recording rules; limit vendor reuse of caller audio; protect SMS content with personal data.","caveats":"Two-day trial journalism; no misroute statistics; assurances are operator statements; older than last edition without a post-trial measured update.","streamIds":["emergency-services"],"roles":{"sales":"Frame Lezle as a monitored non-emergency triage agent that frees thin overnight/dispatcher crews for 911 and radio—never as an autonomous 911 call taker. Discovery should document concurrent workload (radio + 911 + non-emergency), overnight coverage plans, and how the center proves emergency escalation works. Offer a short shadowed trial with human monitoring and public FAQ, not a permanent no-human non-emergency line. Unsupported claims include measured wait-time reduction, zero misroutes, or safe unsupervised overnight AI from this two-day KAIT trial report.","engineering":"Fit is a non-emergency conversational agent with forced escalation to a human when the dialog indicates an emergency, plus post-call SMS for administrative tasks such as fingerprinting. Integrate with contact-center queues, recording, and supervisor barge-in; keep 911 trunks unreachable by the agent. Prerequisites include escalation phrase tests, overnight monitoring staffing, and clear CAD/notes when AI handled a call. Proof of value is critical-escalation success rate, abandoned-call rate, and supervisor intervention frequency during the trial—not anecdotal stress relief.","delivery":"Run a time-boxed trial with live dispatcher monitoring, document every escalation, update SOPs that AI never answers 911, and brief police/fire user agencies on scope. Ownership is the E-911 training/operations lead for go/no-go; QA owns sampling; leadership owns public messaging after concerns. SkillsAdoption teaches monitoring and reclaim, not deskilling emergency answering. AcceptanceCriteria include zero AI-answered 911 calls in logs and successful emergency handoffs in test scripts. Risks include overnight under-monitoring, public distrust after prior AI concern stories, and mission creep beyond non-emergency."},"roleDetails":{"sales":{"customerProblem":"Small PSAP crews juggle radio, 911, and non-emergency administrative calls, creating stress and risk that true emergencies wait behind routine requests.","stakeholders":["E-911 director / training coordinator","Telecommunicators","Supported fire and police chiefs","City/county public information"],"discoveryQuestions":["What is overnight monitoring coverage if AI runs after hours?","How is emergency escalation tested?","How will the public be told AI may answer non-emergency lines?","Which metrics end or extend the trial?"],"valueHypothesis":"A monitored non-emergency AI agent may absorb administrative calls if emergency handoffs are reliable—without claiming wait-time or safety gains unmeasured in this trial story.","potentialEngagement":"Shadowed two-week non-emergency trial design with scripted emergency injects and public FAQ.","unsupportedClaims":["Proven reduction in 911 wait times","Safe unsupervised overnight AI","Zero misroute guarantee","Replacement of dispatcher FTEs"]},"engineering":{"fit":"Non-emergency conversational agent with mandatory human escalation path and supervisor monitoring—not 911 call-taking.","architectureIntegration":"Non-emergency trunk → AI dialog → human queue on emergency cues; SMS gateway; recording and monitoring consoles for dispatchers.","prerequisites":["Escalation test corpus","Always-on barge-in","CAD/annotation of AI-handled calls","Staff schedule for monitoring"],"constraints":["Short public trial evidence only","Administrative examples may not generalize to complex non-emergency public-safety calls","Prior local AI concern history may constrain acceptance"],"security":"Secure voice/SMS PII; vendor limits on model training; audit logs of escalations and overrides.","proofOfValue":["Emergency escalation success on injected tests","Supervisor intervention rate","Non-emergency handle time versus baseline"]},"delivery":{"work":"Execute monitored trial, log escalations, update SOPs forbidding 911 AI answering, brief served agencies, and decide overnight go/no-go with QA evidence.","dependencies":["Vendor configuration access","Dispatcher overtime/monitoring","Legal/PIO messaging","Radio/911 staffing floor"],"ownership":"Training/operations coordinator owns trial control; director owns production decision; QA owns sampling.","skillsAdoption":"Train staff to monitor, reclaim, and document AI sessions without losing emergency call-taking proficiency.","governanceCheckpoints":["End-of-trial review","Overnight monitoring approval","Public communications before expansion"],"acceptanceCriteria":["No AI-answered 911 calls in logs","Successful emergency handoffs in tests","Written SOP with never-answer-911 rule"],"risks":["Under-monitored overnight operation","Public backlash","Scope creep beyond non-emergency"]}},"retrievedAt":"2026-09-15T18:49:19Z","enrichedAt":"2026-09-15T18:52:03Z","enrichmentBasis":"retrieved source","accessibilityWorkforceImplications":"AI may ease trainee/dispatcher load but training and monitoring still require experienced telecommunicators; SMS follow-ups must remain usable for callers without smartphones via human fallback.","procurementImplications":"Structure purchase as a trial with exit rights tied to escalation QA; do not fund FTE reductions from a two-day demo.","operatingModelImplications":"Keep AI on non-emergency only with live human oversight; maintain minimum staffing for 911 and radio even when AI is active.","sourceVerification":{"openedUrl":"https://www.kait8.com/2026/09/02/jonesboro-dispatch-center-is-testing-an-ai-call-taker/","referenceExcerpt":"She's never going to answer a 911 call","promptVersion":"sled-research-v4.0","model":null,"basis":"agent-reported inspection"}}}]}