Loading...
Loading...
Ohio's state government has produced two of the most compelling government AI ROI cases in the country in the past five years, and both are worth understanding before any agency or municipality in the state evaluates a new deployment. The Ohio Bureau of Workers' Compensation ran a machine learning fraud detection program that returned documented net recoveries of $45 million over a multi-year period — a case study that the National Association of State Chief Information Officers has cited as a benchmark and that has set an expectation among Ohio agency directors that AI fraud tools should be measured against hard dollar outcomes, not efficiency metrics. The second landmark case is the ongoing Medicaid NextGen rebuild: the Ohio Department of Medicaid partnered with DXC Technology to replace the state's legacy MMIS (Medicaid Management Information System) with a modern platform, a project that is one of the largest and most technically complex state government IT undertakings in the country and that is generating demand for AI-enhanced eligibility verification, prior authorization automation, and provider analytics that will run for years. Outside Columbus, Wright-Patterson Air Force Base in Dayton is home to the Air Force Research Laboratory, which employs thousands of scientists and engineers working on AI-enabled systems — and generates a civilian and contractor workforce with direct AI development experience that makes the Dayton-Springfield area a deeper AI talent market than its population size suggests. Columbus's Smart City ambitions, funded in part by a $40 million U.S. Department of Transportation Smart City Challenge grant, have created a municipal government with more data infrastructure and AI appetite than almost any comparable mid-American city. LocalAISource connects Ohio agencies with practitioners who have navigated the BWC and Medicaid benchmarks that Ohio decision-makers will hold every new proposal against.
Updated June 2026
The Ohio Bureau of Workers' Compensation fraud detection program is the most cited government AI case study in the Midwest, and understanding what made it work explains what Ohio agencies will demand from any new AI fraud engagement. The program combined claim pattern analysis, provider billing anomaly detection, and social network analysis of related-party schemes — providers billing for treatments never rendered, injured workers claiming for pre-existing conditions, and employer premium classification fraud. The ML models were trained on BWC's own historical claims data, which is among the richest workers' compensation datasets in the country given Ohio's large manufacturing and construction workforce. The $45 million figure represents net recoveries after model development, investigation support, and prosecution costs — it is not gross flagged amounts but actual money recovered. That distinction matters because Ohio auditors, particularly the State Auditor's office under David Yost's prior leadership and now his successor, hold AI program evaluations to outcome-based standards. For agencies considering new AI fraud investments, the BWC benchmark creates a practical shortlist criterion: consultants who cannot cite comparable workers' comp, Medicaid, or unemployment insurance fraud ROI data from comparable-scale deployments elsewhere are starting from behind in Ohio's procurement conversations. The Ohio Department of Job and Family Services, which administers SNAP and TANF alongside Medicaid, has piloted similar pattern-analysis tools for benefits fraud since the BWC program demonstrated the approach, with initial results showing a 12-to-1 return on fraud investigation prioritization efficiency.
The Ohio Department of Medicaid's NextGen MMIS rebuild, the most significant state IT investment in recent Ohio history, is converting a decades-old claims processing system into a cloud-native modular Medicaid enterprise system with DXC Technology as the primary integration partner. The new platform creates a data infrastructure capable of supporting AI-enhanced functions that the legacy system could not: real-time eligibility determination, predictive care management outreach, and ML-based prior authorization screening that reduces the average 7-day prior auth turnaround for complex medical procedures. Ohio Medicaid covers more than 3 million residents — over a quarter of the state's population — and the prior authorization backlog has been a persistent political issue, particularly for behavioral health services. AI-assisted prior auth is the highest-visibility near-term deliverable from the NextGen platform, and the ODM has been explicit about expecting DXC and third-party AI vendors to demonstrate behavioral health PA automation capability by 2025. For county-level jobs and family services agencies — Ohio has 88 counties, each with its own DJFS operation — the NextGen platform creates standardized data interfaces that, for the first time, allow county-level AI tools to connect to state Medicaid data without custom integration work. This is a significant unlock for the state's largest counties — Cuyahoga, Franklin, and Hamilton — which have the volume to justify county-level AI investments in case prioritization and client communication that smaller counties cannot support. Columbus is additionally running a parallel smart city AI program through the Columbus Partnership and Smart Columbus initiative, with AI-powered transit optimization, infrastructure monitoring, and resident services portals that have become a practical testbed for municipal AI tools that other Ohio cities observe before making their own procurement decisions.
Wright-Patterson Air Force Base's Air Force Research Laboratory is the largest research enterprise within the Air Force and employs more PhDs than most universities in the country, with substantial concentration in AI, machine learning, autonomy, and human-systems integration research. The civilian and contractor workforce supporting AFRL — which includes major defense technology firms like L3Harris, Leidos, SAIC, and Booz Allen Hamilton with Dayton offices — creates an AI talent concentration in the Dayton-Springfield area that shapes what Ohio government agencies can realistically recruit and contract. Several senior Ohio state IT leaders have AFRL-adjacent backgrounds, and AFRL's rigorous systems engineering culture influences how Ohio agencies evaluate AI vendor proposals — they are used to seeing formal verification and validation documentation, not just vendor-supplied accuracy metrics. The Ohio Supercomputer Center at Ohio State University provides state agencies with high-performance computing access and data science expertise, and has partnered with agencies on AI development projects where internal capacity is insufficient. Government AI engagements in Ohio run $150,000 to $900,000 for scoped state agency deployments — higher than Midwest averages because of the BWC and Medicaid benchmarks that drive rigorous outcome documentation requirements, which add evaluation and audit cost to every project. We've seen a consistent pattern in Ohio government AI engagements: proposals that lead with efficiency savings rather than fraud recovery or benefits-access improvement outcomes get significantly less traction with Ohio agency directors, who have been schooled by the BWC case to think in terms of dollar-denominated outcomes.
Strategic planning for AI adoption, readiness assessment, and roadmap development
Workflow automation using AI, including Make.com-style automation and RPA
Predictive models, data analysis, and ML pipeline development
Text analysis, document automation, sentiment analysis, and language processing
The BWC program worked because it combined three inputs that other agencies often lack: a large, clean, internally consistent claims history going back 15-plus years; an analytics team with actuarial experience who could translate model flags into investigation priorities; and a prosecution partnership with the Ohio Attorney General's Special Investigations Unit. Agencies that want comparable ROI need all three components — ML models alone, without the investigation and prosecution infrastructure, generate flags that never convert to recoveries. The Ohio Department of Job and Family Services and the Ohio Department of Medicaid have both piloted versions of this approach, and early results suggest the model is replicable for benefits fraud with 18 to 24 months of model training time.
DXC Technology's NextGen rollout has been phased, with core claims processing and eligibility infrastructure prioritized before AI add-ons. Prior authorization automation for behavioral health services is on the ODM's 2025 implementation roadmap, with standard medical PA automation following in 2026. Third-party AI vendors can integrate with the NextGen platform through FHIR-compliant APIs that DXC has published, and the ODM has an active vendor evaluation process for PA and care management AI tools. Counties with the most acute PA backlog — Cuyahoga, Franklin, and Summit — are piloting county-level interim tools while the state platform matures.
The Smart Columbus initiative generated open datasets, vendor evaluation documentation, and implementation case studies that the Columbus Partnership has made available to other Ohio municipalities. Cities like Dayton, Toledo, and Cleveland have used Columbus's transit AI and infrastructure monitoring pilots as references when writing their own RFPs, saving significant discovery time. The Ohio Municipal League facilitates peer learning between cities on technology adoption, and Smart Columbus staff have presented at OML conferences. The practical benefit for smaller municipalities is a vetted vendor list and realistic performance benchmarks rather than relying entirely on vendor-supplied claims.
AFRL's technology transfer office, under 711th Human Performance Wing and AFRL's Information Directorate, has active agreements with Ohio state agencies and universities for dual-use AI tools — particularly in autonomous inspection (applicable to ODOT infrastructure monitoring), natural language document processing (applicable to state licensing and permitting), and human-machine teaming tools for decision support. The Ohio State University's Translational Data Analytics Institute serves as a bridge between AFRL research and civilian agency deployment. Formal technology transfer typically takes 12 to 24 months, but the talent network effect — AFRL researchers consulting with state agencies informally — operates much faster.
A county DJFS AI pilot — typically focused on case prioritization, document intake automation, or client communication — runs $60,000 to $180,000 in year one for a mid-sized county like Stark, Montgomery, or Lucas. Franklin and Cuyahoga, with higher case volumes, run $150,000 to $300,000. The NextGen MMIS data interfaces reduce integration cost significantly for counties that have completed their NextGen data migration, which most large counties will have finished by late 2025. Timeline from contract to production for a scoped pilot is typically 4 to 6 months; the Ohio Office of Budget and Management's IT investment review adds 30 to 60 days to procurement for projects over $500,000.
Get found by Ohio businesses on LocalAISource.