{"resourceId":"suffolk-county-ny-cora-ai-voice-agent-consumer-affairs-comptroller-2026","versions":[{"version":"external-711965508670f3187db5af405c755d60001ffc814519e5fee002a911960a2522","resource":{"id":"suffolk-county-ny-cora-ai-voice-agent-consumer-affairs-comptroller-2026","title":"Suffolk County, New York's CORA voice agent: vendor reports 45-52% autonomous resolution and live tax-due lookups","organization":"Suffolk County, New York (Department of Information Technology)","sector":"County contact center (Consumer Affairs, Comptroller)","geography":"Suffolk County, New York, United States","publishedAt":"Undated Google Cloud customer page, retrieved 26 October 2026; listed by a third-party use-case index with a 23 October 2026 evidence date","publicationDate":null,"eventDate":null,"sourceName":"Google Cloud","sourceLabel":"Vendor customer case study quoting county IT","sourceUrl":"https://cloud.google.com/customers/suffolk-county-new-york","evidenceClass":"vendor-claim","outcomeClass":"emerging","topics":["developers-agents","operating-model","data-security","infrastructure","accessibility-workforce"],"finding":"Google Cloud reports that Suffolk County's CORA voice agent, built by UVS InfoTech, resolves 45-52% of Consumer Affairs calls without staff on up to 550 calls a week, using live lookups in Salesforce and Accela. In the Comptroller's Office it reads back property-tax amounts due from the 10 towns' portals, handled 874 calls in two weeks, resolved 36% directly, and is credited with a 30% drop in incoming calls during tax season. Baselines cited include waits over 30 minutes at peak. These are vendor-published figures without measurement method or error rates.","sledRelevance":"Why now: surfaced in late October 2026 (page undated). One of the more specific county examples of an AI agent doing live record lookups and calculations rather than FAQ answers, which raises accuracy and liability questions other counties will face.","evidence":"Vendor case study figures (resolution rate, call counts, 30% volume reduction, baselines) and county IT quotes; no independent evaluation.","architectureImplications":"Contact-center AI on a hyperscaler platform integrated with existing telephony, CRM (Salesforce), permitting/licensing (Accela) and 10 separate town tax portals; telemetry feeds a dashboard.","governanceImplications":"An agent that reads back amounts owed including penalties needs accuracy testing, audit logs and a clear policy on reliance; none are described.","securityPrivacyImplications":"Tax function described as ring-fenced from internal CRM; Consumer Affairs function reads and writes CRM records, so access controls and logging matter.","caveats":"Vendor-published and undated; no definitions of 'resolution'; no error rate for tax figures; synthetic training data; volume drop may be seasonal or channel shift.","streamIds":["local-government"],"roles":{"sales":"Framing: counties are moving from FAQ bots to agents that look up records. Google Cloud reports Suffolk County's CORA resolves 45-52% of Consumer Affairs calls and reads back tax amounts due from town portals. Interpretation: this supports discovery with counties facing seasonal call surges. The figures are vendor-published without method, so they cannot be quoted as typical results, and any proposal should price in accuracy testing for calculated amounts.","engineering":"Framing: the integration surface is the real project. CORA connects telephony, Salesforce, Accela and 10 separate town tax portals. Interpretation: a comparable build needs stable APIs or reliable portal access for each system, a clean handoff with transcript, and per-function data segmentation like the ring-fencing described. Using synthetic data to expand training FAQs means answer quality must be tested against real call samples before go-live.","delivery":"Framing: resolution rate is only meaningful with a definition. Suffolk reports 45-52% autonomous resolution and a 30% call drop, but not how either was measured. Interpretation: a delivery plan should define resolution, sample calls for correctness, especially tax amounts, compare volumes with the prior tax season, and track repeat calls that suggest unresolved issues counted as resolved."},"roleDetails":{"sales":{"customerProblem":"Seasonal call surges overwhelm limited county staff, producing long waits and manual escalations.","stakeholders":["County IT","Comptroller or treasurer","Consumer affairs","Town tax receivers","Language access office"],"discoveryQuestions":["Which call types peak seasonally?","Which systems would an agent need to read or write?","How would a wrong amount quoted to a resident be handled?","How is resolution measured today?"],"valueHypothesis":"Bounded: an agent with live lookups may absorb part of routine call volume, with the share to be measured locally.","potentialEngagement":"A call-type analysis and accuracy test plan for one department before a voice-agent pilot.","unsupportedClaims":["45-52% resolution as a typical result","Accuracy of tax amounts","Staff cost savings","Resident satisfaction"]},"engineering":{"fit":"Counties with CRM and licensing platforms that expose APIs and high-volume routine call types.","architectureIntegration":"Contact-center AI on existing telephony, with live reads and writes to CRM and licensing systems and lookups against external tax portals.","prerequisites":["System APIs or reliable portal access","Call-type data","Transcript handoff to staff queues"],"constraints":["Fragmented town-level systems","Hyperscaler platform dependence","Language quality varies"],"security":"Segment functions so the tax lookup cannot reach internal CRM, log all record access, and authenticate callers before sharing account data.","proofOfValue":["Sample calls to verify quoted tax amounts","Define and audit 'resolution' on a call sample","Compare call volume with the prior season"]},"delivery":{"work":"Deploy an AI voice agent for routine county calls with measured resolution and accuracy.","dependencies":["Department system owners","Telephony provider","Systems integrator"],"ownership":"County IT, with each department owning its call content and accuracy.","skillsAdoption":"Agents handle escalations with transcripts; IT monitors telemetry dashboards.","governanceCheckpoints":["Accuracy sign-off for calculated amounts","Language access review","Quarterly review of resolution definitions and repeat calls"],"acceptanceCriteria":["Proposed: resolution defined in writing and audited monthly on a call sample","Proposed: quoted amounts match records in a sampled test","Proposed: repeat-call rate tracked against baseline"],"risks":["Wrong tax amounts quoted at deadlines","Inflated resolution metrics","Vendor and platform lock-in"]}},"retrievedAt":"2026-10-26T22:17:52-05:00","enrichedAt":"2026-10-26T22:22:49-05:00","enrichmentBasis":"retrieved source","accessibilityWorkforceImplications":"Seven languages aligned with the county's language access executive order; after-hours coverage; staff described as freed for complex cases, with no staffing data.","procurementImplications":"Delivered by an 8(a) systems integrator on a hyperscaler platform; contract value not disclosed; planned ERP and payment expansion would deepen dependence.","operatingModelImplications":"Replaces 25-plus IVR branches with intent-based routing and transcript handoff; staff work shifts to escalations and field work.","sourceVerification":{"openedUrl":"https://cloud.google.com/customers/suffolk-county-new-york","referenceExcerpt":"CORA consistently maintains an autonomous resolution rate of 45–52%, managing up to 550 calls per week without human intervention.","promptVersion":"sled-research-v4.0","model":null,"basis":"agent-reported inspection"}}}]}