Loading...
Loading...
Indiana state government is navigating three simultaneous AI investment cycles that rarely overlap but all draw from the same limited pool of public-sector technology talent. The Indiana Family and Social Services Administration (FSSA) operates the Curam-based KyCares eligibility system — a federally-certified IBM platform supporting Medicaid, SNAP, TANF, and CHIP for 1.7 million Hoosiers — and is under CMS pressure to add predictive analytics and fraud detection capabilities without destabilizing the eligibility logic that governs $14 billion in annual Medicaid expenditures. Across state government, the Indiana Department of Environmental Management (IDEM) is processing an unprecedented volume of Title V and NPDES permit applications driven by Eli Lilly's $9 billion manufacturing expansion across Lebanon, Branchburg, and Indianapolis — expansion-related permit applications that require complex air dispersion modeling and cumulative impact analysis that IDEM's existing staff cannot process at pace. And in Indianapolis, the city's Office of Public Innovation and Improvement (OPII) has been the most active local government AI adopter in the state, piloting predictive code enforcement, AI-assisted 311 routing, and a council-district-level performance analytics dashboard that has become a reference model for mid-size Midwest cities. LocalAISource connects Indiana government agencies, IDEM-regulated industries, and Indianapolis city departments with AI professionals who understand the specific platforms, regulatory frameworks, and political dynamics that shape public-sector technology investment in this state.
Updated June 2026
Indiana's FSSA KyCares system runs on IBM Curam 7, a platform that handles eligibility determination, case management, and benefits issuance for approximately 1.7 million Medicaid enrollees plus additional populations in SNAP, TANF, and childcare assistance programs. CMS-certified eligibility systems are governed by the Medicaid Enterprise Certification Toolkit (MECT), which constrains the architecture of any AI layer built on top: changes to the eligibility rules engine require APD (Advanced Planning Document) approval from CMS, a process that can take 6 to 18 months. This means AI work at FSSA is primarily analytics-adjacent rather than embedded in the core eligibility determination — predictive risk stratification that identifies members at high risk of chronic disease progression, prior-authorization NLP that routes requests to the correct clinical reviewer, and FWA detection on provider claims. The Indiana Medicaid FWA program, administered through FSSA's Office of Inspector General in coordination with the Indiana Medicaid Fraud Control Unit (MFCU), processed $78 million in fraud recoveries in fiscal year 2024. The current FWA detection system uses a rules-based approach on the Gainwell Technologies claims processing platform; the modernization case for ML-based anomaly detection is strong, but the FSSA OIG has been cautious about false-positive rates after an earlier rules tightening in 2022 generated a wave of provider complaints that created congressional attention. Any FWA AI vendor must demonstrate calibrated recall-precision tradeoffs at the Indiana provider-mix level, not just national averages.
Eli Lilly's announced $9 billion Indiana manufacturing expansion — covering the Lebanon manufacturing campus (API synthesis), the Branchburg biologics site, and expanded operations at the Indianapolis Corporate Center — represents the largest single-employer environmental permitting load IDEM has faced in a generation. Each new manufacturing building requires Title V major source air permits, NPDES stormwater and process-water permits, and in some cases RCRA hazardous waste determinations. IDEM's Office of Air Quality has a target of 180-day permit processing time for major sources; the Lilly expansion has pushed queue depth well beyond that, and the agency received additional appropriations in the 2024 supplemental budget specifically to hire permit engineers. AI-assisted permit review is a direct response to this pressure. The technology suite in play includes: NLP extraction from permit applications to auto-populate modeling input templates (cutting data entry time per application from 8 hours to under 2), air dispersion model output comparison tools that flag where applicant-submitted AERMOD results deviate from IDEM's internal modeling re-runs, and cumulative impact analysis tools that aggregate emissions from multiple Lilly facilities against the EPA's EJScreen environmental justice indicators for surrounding communities. IDEM is also a participant in the EPA's ECHO (Enforcement and Compliance History Online) data integration program, and AI tools that cross-reference self-reported emissions against ECHO compliance history can prioritize permit reviews for applicants with prior violation patterns. In practice, the gap between a permit that sails through in 120 days and one that takes 300 days is often determined by application completeness and emissions modeling quality — AI pre-screening on the applicant side, not just the IDEM side, has measurable value for Lilly's project timeline.
Indianapolis's Office of Public Innovation and Improvement has been the most visible municipal AI adopter in Indiana for the past three years. OPII's predictive code enforcement model — which uses property characteristic data, permit history, utility payment patterns, and 311 complaint history to identify properties at elevated risk of code violations before complaint-driven inspection — has been cited by the Urban Institute and the What Works Cities network as a best-practice example of proactive government service delivery. The model runs quarterly and generates a priority inspection list that OPII shares with the Department of Business and Neighborhood Services (BNS). The AI-assisted 311 routing system, which OPII built on top of the city's Salesforce Government Cloud implementation, reduces misrouted service requests by classifying natural-language descriptions of problems into the correct city department intake queue. Before deployment, approximately 22% of 311 requests were routed to the wrong department; post-deployment that figure dropped below 8%. Both the 311 routing and code enforcement models were developed through a partnership with the Regenstrief Institute — Indiana University's health informatics research center, which has become an informal technical partner to Indianapolis city government on data science projects because of its proximity and public-sector mission alignment. For other Indiana municipalities evaluating AI investments, OPII's work is a double-edged reference: the methodology is rigorous and replicable, but the technology lift — Salesforce Government Cloud plus custom ML models plus the data pipeline infrastructure — represents a $2.5 million to $5 million investment that Fort Wayne, Evansville, and South Bend cannot replicate without a different implementation approach. Lighter-weight SaaS alternatives from vendors like SeeClickFix, Motorola Solutions, and Tyler Technologies serve smaller Indiana municipalities at $80,000 to $250,000 annually.
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
Yes, but the architecture must separate AI analytics from the core eligibility rules engine. CMS requires that any change to eligibility determination logic go through the APD approval process, which takes 6 to 18 months. The compliant approach is to build AI analytics as a read-only layer against a replicated copy of KyCares data — a data warehouse or lakehouse architecture that receives claims and case data but cannot write back to the production eligibility system. Risk stratification and FWA detection are the primary use cases in this architecture. FSSA's existing data warehouse on the Gainwell Technologies platform provides the extraction point; the AI layer runs on top of extracted data, not the live Curam instance.
The Lilly expansion has created an acute operational pressure that is accelerating IDEM's AI evaluation timeline. The agency received supplemental budget appropriations in 2024 specifically for permit processing capacity, and AI-assisted review tools are part of that capacity plan rather than a separate innovation initiative. Vendors with experience in EPA AERMOD integration, Title V permit NLP, and EJScreen data overlay have a clear RFP fit with IDEM's immediate needs. The entry point is typically a pilot scope covering one permit category — major source Title V or NPDES — at $200,000 to $500,000, before expansion to agency-wide deployment.
The core data ingredients — parcel records, permit history, utility payment data, and 311 complaint logs — are available in most Indiana cities with populations above 50,000, because Marion County's data infrastructure is not uniquely rich. What OPII had that other cities lacked was the Regenstrief Institute partnership, which provided data science capacity at lower cost than commercial consulting. Fort Wayne, Evansville, and South Bend can replicate the methodology by partnering with Purdue University, Indiana University, or University of Notre Dame data science programs, which have municipal government engagement programs. The technology cost for a smaller city is $150,000 to $400,000 for initial model development, plus ongoing model refresh costs of $50,000 to $100,000 annually.
Indiana's MFCU, housed within the Attorney General's office, coordinates with FSSA OIG on fraud referrals but conducts its own investigations with a focus on criminal prosecution rather than civil recovery. The MFCU has used ML clustering analysis on provider billing patterns to identify case development priorities since 2021 — specifically, cluster analysis of billing-code combinations that are statistically implausible given patient diagnosis codes. In 2024, the MFCU obtained convictions in 14 Medicaid fraud cases, several of which originated from ML-flagged billing anomalies. The MFCU uses the OIG's national Medicaid data analytics platform (T-MSIS Analytic Files) as its data source, so Indiana-specific ML work supplements rather than replaces federally provided analytics tools.
The Indiana Office of Technology (IOT) maintains a state technology contract library that includes several pre-approved AI and analytics vendors — primarily through the Salesforce, Microsoft Azure, and AWS state enterprise agreements. Municipalities can piggyback on state contracts under Indiana's cooperative purchasing statute (IC 5-22-17), which eliminates the need for an independent RFP and can cut procurement timelines from 12 months to 60 to 90 days. IOT also administers the Statewide IT Solutions (SITS) contract vehicle, which includes AI and data analytics services. For smaller municipalities without IT staff, the Indiana Fiscal and Management Analysis office (IFMA) provides free technology procurement consulting.
Get your practice listed on LocalAISource today.
Get Listed