Municipal governments are under continued pressure to save money while delivering more and better services, noted John Dvorak, chief technology officer for public sector at Red Hat, during the recent Red Hat Government Symposium.
In Colorado, the City and County of Denver is addressing those pressures with automation – and putting hard numbers behind its efforts. The Red Hat Government Symposium session, “Data to Decisions: Scaling Automation Across Government,” explored those numbers and the efforts that led to measurable success.
Holly Troy, senior DevOps automation architect for Denver, said the city and county automation capability started small and scaled through a deliberate strategy: “We want humans to be transformative and automation to be transactional.”
In other words, Denver government is using automation to handle routine tasks and free staff time for higher-value work. Initially, the effort focused on IT use cases such as server builds. Then, it expanded through socializing the opportunity with other departments and enabling workers to submit automation requests via ServiceNow.
A key part of Denver’s approach is prioritization and measurement. A return on investment scoring model, called RICE, is used to evaluate reach, impact, confidence, and effort before work begins.
In other words, Troy said, “How many times a month is a process going to be done, and what kind of impact is it going to have, either internally or externally facing? … Are we confident that we’re going to be able to get this automation completed? … How much effort is it going to take?”
Denver moved from “hours saved” to “hours refocused”
Denver tracks automation outcomes in a Microsoft Power BI dashboard that management reviews regularly, Troy said. In 2025, automation saved roughly $2.2 million, he noted.
“But the thing that our management really likes to look at is the number of hours refocused,” he observed. That’s 50,000 hours annually, or 131 hours per day.
That focus on “hours refocused” is tied to how city and county leaders talk about automation with employees – not as a replacement for them, but as a recovery of their human capacity.
For example, near-real-time automation in Denver’s public defender’s office means that multiple staff members no longer need to monitor an inbox and process requests for public defenders. That automation saves staff nine hours per day, and it means that a public defender can meet with a judge about a case within 15 minutes of a request.
Staff members “can actually focus on helping the individuals, so it’s been extremely impactful,” Troy said.
Services are faster, inside and out, and humans are in the loop
Troy emphasized that service improvements often depend on blending automation with human decision points, across both constituent-facing and internal processes.
For example, Denver is streamlining permitting workflows – a pain point for many municipal governments. With Ansible Automation, a construction permit can now be completed within 24 hours instead of three months – but “you still have humans in the loop,” Troy said.
For example, during the permit review process, “Ansible Automation calls AI to validate a driver’s license … and if something’s off, it grabs a human,” he said.
Automation also reduced staff wait time and handoffs in the IT department. One workflow helped technicians working on a domain migration avoid multi-hour delays tied to the release of IP addresses assigned to laptops. “By giving them a form in Ansible, they were able to enter the laptop machine name and release the address” without waiting for the network team, Troy said.
Some improvements are digital, and sometimes they’re shovels
Some of Denver’s most tangible examples are citizen-centered automations that don’t sound like IT modernization at first glance. For example, after a snowfall, the Denver Snow Angels program matches residents 60 years and older who need help shoveling sidewalks with volunteers in the neighborhood.
“That got such a huge response,” Troy said. “It actually hit local news when we first launched it last year, and since then, it’s just gotten faster.”
Early wins like this one create demand for automation across departments. Quite simply, “It sells itself,” Troy said.
Clean inputs and strong guardrails on data are the foundation
Data preparation is a core consideration when building solutions that incorporate automation and artificial intelligence (AI), Holly and Dvorak observed. For example, employees or citizens using a chatbot could inadvertently expose sensitive data such as passwords or personally identifiable information (PII). “You have to put guardrails around the data,” Troy stressed.
In government, sensitivity varies by dataset, and Troy flagged Criminal Justice Information Services (CJIS) data in particular. Agencies must decide how to handle it, and automation can help enforce rules around CJIS or any other class of data, he noted.
“You can use automation such as Ansible to help clean up that data. If you’re expecting a certain length or type of response [and] it’s outside these parameters, let’s either block that or get a human in the loop to make sure we’re not off the rails,” he said.
Troy described using Proofpoint and building additional monitoring to reduce the risk of sensitive data leakage into public AI tools.
“I’m currently working on an automation we’re building out a dashboard in Power BI to capture when PII information is pasted into, for example, ChatGPT,” Troy said. “We want to be alerted when that happens – not necessarily to punish the individual. It’s a training process. … You can’t stop [people] from using it, so you have to teach them how to use it.”
To start, “Find those eagles and let them loose”
Dvorak framed Denver’s progress as more than tooling – a cultural shift also enables adoption. Troy agreed, emphasizing the need to address staff questions and hesitations early and directly.
He builds trust by embedding with departments and learning their workflows end to end, telling staff, “I may not completely understand everything that you do … but I will be [the] best intern you’ve ever had, because I will know this process in and out.”
For other government IT leaders working to create momentum for automation and AI, Troy offered direct advice: Find the person or people who want to take the first steps. “Find those eagles and let them loose. Give them the time they need to do the automations. … And I’m happy to talk to anybody who’s willing to listen.”
Watch the Red Hat Government Symposium session: “Data to Decisions: Scaling Automation Across Government,” and explore more sessions from the Red Hat Government Symposium.