Shawano County, Wisconsin, has just provided a masterclass in smart AI procurement for government. Instead of renewing a pricey all-staff subscription after a six-month pilot, the county opted for a per-user licensing model that will serve only the employees who actually use the tool — slashing its expected costs by roughly 62% and debunking the myth that every desk needs a copilot.
What Actually Happened
In the first half of 2026, Shawano County ran a pilot of GovAI, an artificial intelligence platform designed specifically for government workflows. The price tag: $9,500 for six months, covering all county employees regardless of whether they ever logged in. As the pilot wound down, Technology Services Director Matt Hietpas and department heads reached a common-sense conclusion: the tool had merit, but blanket coverage was overkill.
“During the pilot program period, we determined that this type of tool would not be needed for all county employees,” Hietpas told NEW Media Inc. in a July 18 report. GovAI, apparently responsive to the feedback, offered a per-user alternative at approximately $30 per user per month. The county expects to equip 20 to 30 employees across various departments, which pencils out to $600 to $900 monthly — compared with the old flat rate that worked out to about $1,583 per month. Over six months, the shift will save the county between $4,100 and $5,900.
The users who get GovAI access will be those in departments that demonstrated a clear benefit during the trial. Document drafting, data analysis, resource planning, cross-referencing records, and even parsing body-camera footage for law enforcement were cited as potential use cases. But the county will require staff to review all AI-generated output for accuracy, especially any material destined for public records or decision-making.
Meanwhile, Microsoft Copilot remains on the table. Hietpas noted that Copilot’s tight integration with Microsoft 365 — the county already uses email, SharePoint, and other tools — makes it a compelling alternative or complement. GovAI, conversely, was built from the ground up for the public sector, which might offer workflow advantages. For now, the county is keeping GovAI for a select group while it continues evaluating Copilot.
What It Means for You
This local-government decision carries lessons far beyond Wisconsin.
For government IT leaders and department heads: The Shawano experience is a playbook for avoiding vendor lock-in and overprovisioning. Blanket AI subscriptions are easy to sell but hard to justify once actual usage data comes in. Before committing to any enterprise-wide deal, run a pilot of at least a few months. Track which roles adopt the tool, what tasks it accelerates, and where it gathers dust. Only then negotiate — and insist on per-user pricing if the data doesn’t support broad deployment. Shawano’s move to 20–30 licenses instead of all-employee access is not a failure; it’s a course correction that saves money and focuses tools where they produce value.
For Microsoft 365 administrators in public-sector organizations: The Copilot conversation is important. If your agency already lives inside the Microsoft ecosystem, Copilot’s built-in hooks to Outlook, Word, Teams, and SharePoint could reduce training overhead and simplify compliance. But don’t assume Copilot is automatically the right choice. The county’s CTO openly weighed the government-specific pedigree of GovAI against Copilot’s integration. Your procurement should similarly map the tool’s strengths to your actual workflows, not to brand loyalty. And whatever you choose, lock down data handling, permissions, and audit trails before rollout.
For taxpayers and residents: When a local government cuts a software bill by more than half without sacrificing service, that’s good stewardship. But the story also highlights the need for transparency. AI-generated documents used in law enforcement or public decisions must be reviewed. The county’s policy to mandate human oversight sets a standard other jurisdictions should emulate.
For everyday Windows users: The tools being evaluated — GovAI, Microsoft Copilot, ChatGPT — are the same ones you might encounter in your own workplace. The same principles apply: don’t buy a fleet license if only a fraction of the team will use it; test before you commit; and always, always double-check AI output.
How We Got Here
Shawano County’s journey began when it enrolled in GovAI’s pilot program at the start of 2026. The initial agreement provided countywide access at a flat rate of $9,500 per six-month period. During those months, department heads and employees experimented with using AI for routine tasks: drafting correspondence, crunching spreadsheets, scanning for duplicate records, and summarizing videos. According to Hietpas, the feedback was positive but pragmatic. Many employees didn’t find a use case that justified the tool, and managers saw no reason to pay for licenses that would go unused.
When the pilot neared its end, the county faced a choice: renew the blanket subscription at a high cost, switch to a per-user model (if available), or drop the platform entirely. GovAI stepped up by offering per-user pricing — a move likely prompted by the client’s willingness to walk. For about $30 a head per month, the county could retain the service for the handful of employees who genuinely needed it. That cost is comparable to per-user rates for Microsoft Copilot and ChatGPT Enterprise, Hietpas said.
The decision wasn’t a rejection of AI; it was a rejection of wasteful spending. The county’s budget process now requires individual departments to justify the expense by showing operational value, subject to normal approvals. This bottom-up approach ensures that AI adoption is demand-driven, not top-down.
What to Do Now
If you’re responsible for technology decisions in a government agency, municipality, or even a medium-sized business, here’s what you can learn from Shawano County’s example:
-
Run a structured pilot: Launch a time-limited trial with clear success metrics. Don’t just turn on the tool and wait for feedback. Actively measure logins, task completion, and user satisfaction.
-
Survey actual usage, not enthusiasm: After the pilot, separate the “it’s cool” responses from the “it saves me two hours a week” data. Only the latter justifies a license.
-
Negotiate per-user pricing before committing: Any vendor that offers enterprise flat-fee subscriptions should be pushed to offer per-user or tiered pricing once you have data that shows limited demand. If they refuse, be ready to explore alternatives like Microsoft Copilot or even ChatGPT, which already operate on per-user models.
-
Map integrations before choosing a platform: If your organization runs on Microsoft 365, start by evaluating Copilot because it piggybacks on existing identity, compliance, and collaboration infrastructure. However, don’t rule out domain-specific tools like GovAI if your workflows (e.g., law enforcement, public records management) demand features Copilot lacks.
-
Implement strong review policies from day one: Any AI output that could become part of a public record or influence a government action must be reviewed by a qualified human. Draft a formal policy that spells out who reviews what, and how errors are logged and corrected.
-
Budget by department, not by IT: Let departments build AI costs into their own budgets, as Shawano County now does. That creates accountability and prevents IT from being the sole arbiter of who gets access.
Outlook
Shawano County’s decision may be minor in scale, but it signals a maturing approach to AI in the public sector. Rather than chasing headlines about “transforming government with AI,” agencies are getting pragmatic. The county still hasn’t closed the door on Copilot, and it may well adopt it alongside or instead of GovAI in the future. For now, the message is clear: AI is a tool, not a panacea. Buy only what you need, and make sure a human double-checks the work.
As more governments complete their pilots, expect a wave of similar pullbacks from enterprise-wide licensing toward focused, cost-effective deployments. And for Microsoft, Copilot’s success in state and local government may hinge on its ability to demonstrate value not just as a chat interface, but as a deeply integrated assistant that truly saves time in the workflows public servants actually perform.