{"resourceId":"ny-dol-genai-virtual-agent-employer-37pct-nascio-2026","versions":[{"version":"external-4588039a61d6d8fd04513b625be2894a9f63810baa2efcad1bbd173a71967e53","resource":{"id":"ny-dol-genai-virtual-agent-employer-37pct-nascio-2026","title":"NYS DOL GenAI Virtual Agent (NASCIO): handles 37% of employer conversations since Jan 2026","organization":"New York State Department of Labor with NYS Office of Information Technology Services (ITS)","sector":"State labor / employer services contact center / GenAI virtual agents","geography":"New York State — DOL employer services (implemented January 12, 2026)","publishedAt":"2026-08","publicationDate":"2026-08-01","eventDate":"2026-09-30","sourceName":"NASCIO Awards Library / nomination PDF","sourceLabel":"Virtual Agents — NYS DOL GenAI employer services initiative","sourceUrl":"https://www.nascio.org/wp-content/uploads/2026/08/NY_Artificial-Intelligence.pdf","evidenceClass":"public-sector-association","outcomeClass":"effective","topics":["knowledge-work","accessibility-workforce","operating-model","developers-agents"],"finding":"NYS ITS and DOL’s NASCIO Artificial Intelligence nomination describes the state’s first DOL GenAI Virtual Agent for employer services, implemented January 12, 2026, using Google CCAI/Dialogflow CX with partner Egen to answer FAQ-style employer questions and free staff for complex issues. Impact metrics state Virtual Agents have successfully handled 37% of all employer conversations since deployment, with claimed benefits of instant common-question answers and reduced wait times for complex inquiries. Phased UAT and hypercare for existing agents are part of the implementation narrative.","sledRelevance":"Measured state employer-service GenAI deflection (37%) from a primary NASCIO nomination, complementary to tonight’s award-week coverage and Michigan’s metric-light LARA launch. Why now: PDF not in archive; newly relevant as NASCIO Annual Conference award materials circulate with the 2026-09-30 honors announcement—even though NY was not the AI category winner highlighted by GovTech.","evidence":"Downloaded and pdftotext’d NY_Artificial-Intelligence.pdf; confirmed Jan 12, 2026 go-live, Google CCAI/Dialogflow/Egen partnership language, and Key metrics section stating 37% of conversations handled. Also opened the NASCIO award library executive-summary page for Virtual Agents as corroboration.","architectureImplications":"FAQ-grounded contact-center GenAI with phased UAT/hypercare; keep complex employer cases on human agents; instrument conversation containment separately from satisfaction.","governanceImplications":"First GenAI in a DOL business unit needs ongoing accuracy sampling of employer guidance; nomination claims should be tied to monitored knowledge-base freshness.","securityPrivacyImplications":"Employer conversations may include FEINs/business data—apply contact-center data retention, redaction, and no-training clauses.","caveats":"Self-reported award nomination; 37% is containment/handling share, not independently audited accuracy or CSAT; vendor names appear in nomination; eventDate marks award-week relevance.","streamIds":["state-government"],"roles":{"sales":"Frame NYS DOL’s GenAI Virtual Agent as a measured employer FAQ containment story—37% of conversations handled since January 2026—not a guarantee of accuracy or statewide labor-AI readiness. Discovery should ask how containment is defined, what accuracy sampling exists, and who owns FAQ ground truth. Bounded engagement: a contact-center GenAI containment and accuracy baseline on one employer journey. Unsupported claims include 37% accuracy, guaranteed wait-time cuts for all callers, or that Dialogflow is required.","engineering":"Fit is FAQ-grounded contact-center GenAI with human handoff—not adjudicative unemployment decisions. Architecture should connect a curated knowledge corpus to the virtual agent, then to containment analytics and a human queue for complex intents, with business identifiers redacted in logs. Prerequisites include a FAQ inventory, UAT cycles, and hypercare staffing. Constraints: nomination self-report and accuracy unpublished here. Proof of value is containment plus sampled answer accuracy and recontact rate.","delivery":"Delivery tunes intents, refreshes employer FAQs, samples transcripts weekly for accuracy, and keeps hypercare staffing until containment stabilizes under DOL ownership. Dependencies include DOL employer services, ITS, the platform vendor, and contact-center operations. Ownership of answer correctness sits with the DOL business owner. Acceptance criteria: a defined containment metric, a weekly accuracy sample, and a documented escalation path. Risks include stale FAQs driving wrong guidance and mistaking containment for resolution quality."},"roleDetails":{"sales":{"customerProblem":"Employer contact centers need deflection with measurable quality, not chatbot vanity launches.","stakeholders":["DOL employer services","State ITS","Contact-center ops","CIO/CAIO","Procurement"],"discoveryQuestions":["How is the 37% calculated?","What accuracy sample accompanies containment?","Who updates FAQ ground truth?","What is out of bot scope?"],"valueHypothesis":"FAQ-grounded GenAI virtual agents can contain a material share of employer conversations if accuracy sampling and hypercare are funded.","potentialEngagement":"Contact-center GenAI measurement pilot on one employer FAQ domain.","unsupportedClaims":["37% equals answer accuracy","Guaranteed wait-time reduction for all","Vendor platform lock-in required","Safe for adjudicative decisions"]},"engineering":{"fit":"Grounded virtual agent for employer FAQs with human escalation.","architectureIntegration":"FAQ/knowledge store to CCAI/virtual agent to analytics to human agent desktop.","prerequisites":["FAQ corpus","UAT cycles","Analytics export","Data retention rules"],"constraints":["Self-reported metrics","Accuracy not in nomination","Partnered delivery complexity"],"security":"Minimize FEIN/PII in logs; enterprise tenancy; no model training on chats without agreement.","proofOfValue":["Containment rate","Sampled accuracy","Recontact within 72 hours"]},"delivery":{"work":"Launch/iterate intents; sample transcripts; refresh FAQs; maintain hypercare.","dependencies":["DOL business","ITS","Vendor","Contact center"],"ownership":"DOL owns answer correctness; ITS owns platform runtime.","skillsAdoption":"Agent training on handoff cues when bot confidence is low.","governanceCheckpoints":["UAT sign-off","Weekly accuracy sample","FAQ freshness review"],"acceptanceCriteria":["Containment definition documented","Accuracy sample running","Escalation path live"],"risks":["Stale guidance","Containment without resolution","Scope creep into adjudication"]}},"retrievedAt":"2026-09-30T22:21:20-05:00","enrichedAt":"2026-09-30T22:21:20-05:00","enrichmentBasis":"retrieved source","accessibilityWorkforceImplications":"Instant answers can reduce wait burden for employers if channels remain accessible; staff still needed for complex cases—plan workforce roles around the 63% not contained.","procurementImplications":"CCAI/partner delivery should include accuracy SLAs, knowledge refresh obligations, and exportable conversation analytics in the contract.","operatingModelImplications":"Hypercare plus phased releases is the operating model; DOL business ownership of FAQ truth must continue after ITS delivery.","sourceVerification":{"responseId":"curl-pdftotext-ny-dol-va-20260930T222120","openedUrl":"https://www.nascio.org/wp-content/uploads/2026/08/NY_Artificial-Intelligence.pdf","referenceExcerpt":"Virtual Agents have successfully handled 37% of all employer conversations","promptVersion":"sled-research-v4.0","model":null,"basis":"agent-reported inspection"}}}]}