Loading...
Loading...
Vermont runs a smaller state government than almost any other in the country, but it has made a structural bet on centralization that makes it an instructive case for AI in lean government. The Agency of Digital Services — created in 2017 by consolidating IT functions from across state government into a single agency — operates as Vermont's internal technology delivery organization, managing infrastructure, application development, and enterprise systems for nearly all state agencies. That consolidation model means that AI adoption decisions in Vermont are effectively made once, at the ADS level, and propagate across government — a different procurement dynamic from states where each agency operates its own IT independently. Green Mountain Power, the dominant electric utility serving 80% of Vermont's electric customers, has been a nationally recognized leader in grid modernization, deploying battery storage, demand-response programs, and grid analytics in ways that blur the line between utility and public infrastructure AI. GlobalFoundries' semiconductor fab in Essex Junction — Vermont's largest manufacturing employer — has created a security-clearance-aware contractor workforce with FedRAMP expectations that influence broader technology procurement culture in the Burlington area. Vermont's Public Records Act is one of the more demanding open-government statutes in New England, creating transparency obligations for AI-assisted government decisions that agencies must address proactively. And the University of Vermont in Burlington, through its complex systems and data science programs, has been an applied research partner for state agency AI pilots. LocalAISource connects Vermont government entities with AI professionals who understand the ADS consolidated procurement model, Green Mountain Power's grid AI context, and the Public Records Act compliance requirements that apply to algorithmic decision-making in Vermont.
Updated June 2026
Vermont's Agency of Digital Services is unusual among state IT organizations in having genuine authority over technology decisions across agencies — the model is closer to a corporate IT function than to a typical state CIO office that issues guidance but lacks implementation control. For AI vendors, this creates a concentrated procurement environment: rather than navigating individual agency relationships across 20-plus departments, the path to serving Vermont state government runs through ADS. The Agency evaluates AI tools against Vermont's data governance standards, privacy law requirements, and the Public Records Act's transparency provisions before agency deployment. ADS has also established a cloud-first policy that influences which AI platforms are eligible for state procurement — tools must operate in cloud environments that meet Vermont's data residency and security standards, which for sensitive data typically means ADS-reviewed cloud platforms rather than arbitrary commercial SaaS. The Agency has published an AI use policy framework that distinguishes between low-risk automation (document classification, scheduling, routine correspondence) that agencies can deploy with ADS notification, and high-stakes algorithmic decision-making (benefits eligibility, law enforcement, hiring) that requires ADS review and public disclosure. Vermont's Public Records Act applies to automated decision systems — algorithmic determinations affecting citizens' rights or benefits are subject to records requests — and the ADS framework accounts for this by requiring audit logs and plain-language decision explanations for any AI deployed in citizen-facing contexts. We've seen a pattern repeat in Vermont government engagements: the ADS review process adds 45–75 days to AI procurement timelines, but vendors who arrive with ADS-compatible documentation rarely hit substantive rejection.
Green Mountain Power has deployed battery storage in customer homes at a scale and speed that no other U.S. utility has matched — its Bring Your Own Device program, which aggregates customer-owned Tesla Powerwalls and other batteries into a virtual power plant, uses ML load forecasting and grid dispatch algorithms that manage thousands of distributed assets in real time. The grid intelligence infrastructure GMP has built functions as a proof-of-concept for what AI-assisted public infrastructure management can look like, and the state's Public Utility Commission has engaged with GMP's AI methods in rate cases and grid reliability proceedings. For Vermont state agencies engaged in energy efficiency program administration — the Department of Public Service, Efficiency Vermont — GMP's operational AI models are both a reference case and a potential data-sharing resource. The Vermont Department of Public Service regulates GMP and Eversource Vermont, and the PUC's rate proceedings now routinely include AI-assisted load forecasting evidence that commission staff must be equipped to evaluate. The ADS has been in discussions with the Department of Public Service about a shared analytics platform for utility data that would allow state regulators to run independent ML analyses of utility-submitted load data — a capability that would shift Vermont's regulatory posture from reactive to proactive. GlobalFoundries' Essex Junction fab, which produces RF and specialty analog semiconductors, operates under federal export control and classification frameworks that influence how security-sensitive technology procurement is handled in Chittenden County — the fab's FedRAMP-aware contractor community sets technology security norms that spill into municipal and state government technology conversations in the Burlington metro.
Vermont's government AI priorities reflect a state with unusual demographic pressures — the oldest median age in New England, significant rural population, and a healthcare system anchored by the University of Vermont Medical Center in Burlington that serves as the sole academic medical center for a catchment area covering most of Vermont and parts of New York. The Department of Vermont Health Access, which administers Medicaid under the state's Global Commitment to Health waiver, has been an active pilot environment for AI-assisted eligibility management and care coordination. Vermont's all-payer claims database, one of the most complete in the country, provides a rich data environment for ML models designed to identify high-risk Medicaid members for care management intervention. The Vermont Agency of Human Services' integrated eligibility system — which processes applications for Medicaid, SNAP, Reach Up, and other programs through a single portal — is a candidate for AI-assisted document classification and eligibility screening that would reduce manual review time for caseworkers handling 15-plus program types simultaneously. The University of Vermont's complex systems and computational biology programs have produced applied research partnerships with AHS on social determinants of health modeling. For AI strategy at Vermont scale, implementation costs are compressed relative to larger states — a comprehensive AI roadmap for an ADS-level engagement runs $50,000–$100,000, and single-agency automation pilots run $30,000–$80,000. Vermont's small agency footprint means that a single successful pilot can be adapted across agencies through ADS's shared-service model faster than anywhere else in New England.
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