{"resourceId":"fox8-orleans-carbyne-ai-mva-triage-2026","versions":[{"version":"external-6b34f2abc836bf4ab615519e1bc2804f4d76681af817e09a7439adaea78d3206","resource":{"id":"fox8-orleans-carbyne-ai-mva-triage-2026","title":"Orleans Parish 911 uses AI to answer surge motor-vehicle-accident calls near known crashes","organization":"Orleans Parish Communications District (OPCD) / Carbyne (Axon 911)","sector":"911 / PSAP emergency call triage (fire, EMS, and police dispatch center)","geography":"New Orleans / Orleans Parish, Louisiana, United States","publishedAt":"2026-07-16","publicationDate":"2026-07-16","eventDate":"2026-07-16","sourceName":"WVUE Fox 8","sourceLabel":"Independent local television reporting with OPCD executive quotes","sourceUrl":"https://www.fox8live.com/2026/07/17/when-calling-911-center-new-orleans-ai-may-answer-heres-why/","evidenceClass":"independent-reporting","outcomeClass":"mixed","topics":["operating-model","accessibility-workforce","infrastructure","governance-procurement"],"finding":"Fox 8 (published Jul. 16, 2026) reports OPCD uses Carbyne-developed AI to answer 911 callers within ~200 meters of a known motor-vehicle accident when humans are busy, with Fasold claiming ~15–18 of every 20 crash calls handled by AI, first-three-months 100% listening QA with no false positives/negatives, ~1 hour/day staff productivity, ~$600k/year contract including AI—while stating life-saving evidence would be anecdotal and a Tulane professor warns a single miss is life-critical.","sledRelevance":"Why now: not previously archived; newly relevant after 2026-09-15/16 editions focused on non-emergency-only AI—documents U.S. emergency-path (911) AI for redundant crash surges with human transfer for other emergencies and operator-claimed listening QA.","evidence":"Article quotes Fasold on 200m geofence, queue relief, 15–18/20 crash-call share, 100% listen for first three months with no false positives/negatives, +1 hour/day productivity, staffing gaps (42 vs 64 call takers), ~$600k contract now branded Axon 911; Lalka life-safety caution; Fasold says showy life-save examples unavailable.","architectureImplications":"AI voice agent on 911 for geofenced duplicate MVA reports integrated with CAD knowledge of known incidents; non-matching or other emergencies transfer to humans; accent/ASR training called out as a local constraint.","governanceImplications":"Publish scope (duplicate crash only), retain continuous QA beyond initial three months, and do not equate operator-claimed zero false positives with peer-reviewed clinical outcomes.","securityPrivacyImplications":"AI answers live 911 audio—require vendor limits on training use, retention, and Axon/Carbyne contract transparency similar to other PSAP AI disclosures.","caveats":"Local journalism relying on OPCD estimates; no public independent audit table; life-saving claims explicitly anecdotal; policing and fire/EMS share the same PSAP—stream relevance is dispatch capacity and emergency triage controls.","streamIds":["emergency-services"],"roles":{"sales":"Position geofenced duplicate-crash AI as surge capacity for understaffed 911 centers—not as a general AI call-taker or proven mortality reducer. Discovery must separate Fasold’s 15–18/20 and early listening-QA claims from audited ongoing false-negative rates and ask how non-MVA emergencies are forced to humans. Offer a bounded MVA-surge pilot with continuous QA sampling and public notice. Unsupported claims include guaranteed lives saved, permanent zero false positives, or safe expansion beyond known-crash geofences from this Fox 8 story alone.","engineering":"Fit is a 911 AI voice agent triggered by proximity to CAD-known MVAs, with immediate human transfer otherwise. Architecture needs hard geofence rules, CAD incident state, barge-in/transfer, accent robustness, and full-call recording for QA. Prerequisites include three-month supervised listen plans and clear non-MVA exclusion. Constraints: staffing crisis context and ~$600k bundled contract. Proof of value is ongoing false-negative/positive audits, answer-time for cardiac/critical queues, and telecommunicator available minutes—not anecdotal life-save narratives.","delivery":"Deploy only on duplicate-MVA policy, train reclaim/transfer drills, and keep 100% or high-rate QA after go-live. Ownership stays with the ECC director; QA owns stop-rules if false negatives appear. SkillsAdoption preserves human crash and medical call-taking. Governance checkpoints include monthly false-negative reviews and public scope statements. Risks include mission creep beyond geofenced crashes, over-trust in early zero-error claims, and vendor rebranding opacity (Carbyne/Axon)."},"roleDetails":{"sales":{"customerProblem":"PSAPs face crash-call surges and staffing shortages that bury true emergencies in queue without a safe automation boundary.","stakeholders":["ECC/PSAP directors","Fire/EMS chiefs","City IT/procurement","Public information officers"],"discoveryQuestions":["What geofence and CAD triggers are hard-enforced?","What ongoing false-negative audit exists after month three?","How are non-MVA callers forced to humans?","What is published to callers about AI?"],"valueHypothesis":"If duplicate-MVA AI measurably frees telecommunicators for critical queues under continuous QA, agencies may justify a narrow surge tool—without claiming proven lives saved.","potentialEngagement":"Bounded MVA-surge pilot design with QA sampling and public-notice package.","unsupportedClaims":["Guaranteed lives saved","Permanent zero false positives","Safe general 911 AI answering","Clinical outcome improvement from this article"]},"engineering":{"fit":"Geofenced 911 AI for duplicate MVA reports with human transfer for other emergencies.","architectureIntegration":"CAD known-incident geofence → AI voice → transfer/barge-in; recording to QA store.","prerequisites":["CAD incident location quality","ASR accent testing","Transfer SLA","Recording/retention policy"],"constraints":["OPCD-reported estimates only","Bundled ~$600k contract","Local accent training burden"],"security":"Limit vendor training on 911 audio; authenticate CAD triggers; protect caller PII in AI sessions.","proofOfValue":["Ongoing FP/FN audit rates","Critical-queue answer time","Telecommunicator available minutes","Transfer success for non-MVA"]},"delivery":{"work":"Configure geofence policy, run supervised listen period, train transfer drills, publish caller scope.","dependencies":["CAD accuracy","Vendor ASR tuning","QA staffing for call review"],"ownership":"ECC executive director owns go/no-go; QA supervisor owns stop-rules.","skillsAdoption":"Keep humans sharp on crash and medical intake while supervising AI sessions.","governanceCheckpoints":["End of supervised listen","Monthly FN review","Annual contract metrics"],"acceptanceCriteria":["Documented FN/FP rates","Non-MVA transfer tests passed","Public AI-scope notice posted"],"risks":["Mission creep","Over-trust in early zero-error claim","Anecdotal life-save marketing"]}},"retrievedAt":"2026-09-18T03:16:05Z","enrichedAt":"2026-09-18T03:20:28Z","enrichmentBasis":"retrieved source","accessibilityWorkforceImplications":"Claimed telecommunicator stress relief and +1 hour/day productivity; accent/ASR failures could harm LEP or strong-dialect callers if transfer fails.","procurementImplications":"~$600k/year bundled contract now under Axon 911 branding—require clear AI scope, QA deliverables, and training-data limits in the agreement.","operatingModelImplications":"Splits 911 answering between humans and AI for a narrow duplicate-crash class while retaining human control for other emergencies.","sourceVerification":{"openedUrl":"https://www.fox8live.com/2026/07/17/when-calling-911-center-new-orleans-ai-may-answer-heres-why/","referenceExcerpt":"We had no false positives, [and] we had no false negatives","promptVersion":"sled-research-v4.0","model":null,"basis":"agent-reported inspection"}}}]}