Ohio wins award for AI use in unemployment insurance operations
Ohio earned the Merrill Baumgardner Award for its application of AI in unemployment insurance, according to a press release from the Ohio Department of Job and Family Services (ODJFS). The recognition highlights work that moved from concept to daily use inside core operations.
What Ohio implemented
- Multilingual virtual assistants: Self-service support in multiple languages to answer status, eligibility, and process questions, easing call center load and extending access.
- Document processing: AI-assisted intake and classification to reduce manual keying and speed up claims and adjudication workflows.
- Information management bot: Fast retrieval of policies, guidance, and status details so staff spend less time searching and more time resolving cases.
- Policy simplification via generative AI: Plain-language summaries and Q&A that help claimants and staff interpret rules consistently.
"This award reflects the ongoing commitment of the DeWine-Tressel Administration to improve the experience for Ohioans we serve," Director Matt Damschroder said in the release. "The advancements made in our systems have resulted in faster, more reliable, and more accessible services for individuals who rely on us. This recognition reinforces the importance of continuing to deliver solutions that make a meaningful difference for Ohio families."
Why this matters to insurance and operations leaders
Public UI operations mirror private insurance in intake, verification, adjudication, correspondence, and appeals. Ohio's approach shows where AI can reduce friction without breaking compliance or service levels.
- Speed: Shorter cycle times from intake to determination.
- Quality: Fewer manual touches and more consistent decisions.
- Access: Multilingual support that meets customers where they are.
- Cost: Automation absorbs volume spikes without adding headcount.
How to apply this in 90 days
- Map two high-friction workflows (e.g., initial claims intake, identity verification) and quantify volume, handle time, and error/appeal rates.
- Launch a virtual assistant for your top 10 intents (status, documents needed, payment timing) and route edge cases to live agents.
- Pilot document AI on two forms with clear acceptance criteria (extraction accuracy, touch-time, exception rate).
- Stand up an internal policy Q&A bot using retrieval over your current policy library and SOPs, with citations shown on every answer.
- Set a human-in-the-loop protocol for adverse or uncertain decisions and log overrides for learning.
- Run privacy, security, and accessibility checks before scaling. Include redaction, audit trails, and language quality review.
Metrics to track
- Claim/incident cycle time and backlog age distribution.
- First-contact resolution and call/chat deflection rate.
- Document touch-time, exception rate, and resubmission rate.
- Determination accuracy, appeals rate, and rework.
- Overpayment detection and fraud review yield.
- Customer satisfaction and language coverage utilization.
Guardrails that keep AI production-ready
- Data governance: PII redaction, role-based access, and retention aligned to policy.
- Quality: Translation accuracy thresholds and fairness checks across languages.
- Model oversight: Drift monitoring, fallback responses, and versioned prompts.
- Policy control: Source-of-truth retrieval with citations; change logs for every update.
- Vendors: SLAs on accuracy, latency, and uptime; exit clauses and portability.
- Accessibility: WCAG-aligned experiences across voice, web, and mobile.
For broader context on workforce innovation and awards, see the National Association of State Workforce Agencies (NASWA). The ODJFS site above provides updates on the state's program changes as they roll out.
Want playbooks and case studies that mirror these use cases? Explore AI for Insurance for practical implementations across claims, policy, and service operations.
Your membership also unlocks:
