Nearly every enterprise sector is trying to figure out how AI moves from experimentation into real operational value. But one of the more surprising acceleration stories may be happening inside government.
In this AppDevANGLE conversation at Google Cloud Next 2026, I spoke with Chris Hein, Field CTO at Google Public Sector, about how agencies are adopting Gemini for Government, modernizing decades-old workflows, and preparing for what he describes as an emerging “agentic workforce.”
For years, public sector technology adoption has often been characterized as cautious, compliance-heavy, and slower moving than commercial markets. Yet according to Hein, AI may be changing that dynamic quickly.
“We are actually seeing government in the lead,” Hein said. “They are getting in, and they are outpacing what we’re seeing out of the commercial side.”
That is worth paying attention to.
Public Sector AI Starts With Productivity, Then Expands to Mission Outcomes
Many AI rollouts begin with broad transformation language, but the most successful deployments usually start smaller: helping workers do their jobs better today.
Hein explained that Gemini for Government is often introduced first as a secure environment where employees can safely explore how AI improves day-to-day tasks. “We start to try to make sure that they can be more productive,” Hein said. “They can get more done in each one of their days.”
That practical first step matters. Public sector organizations often face staffing shortages, increasing citizen expectations, and expanding workloads. AI does not need to solve everything on day one. It needs to remove friction, accelerate repetitive work, and create confidence.
Once that trust is built, agencies can move toward rethinking larger mission processes rather than simply optimizing old ones.
Safe Adoption Requires Top-Down Support and Bottom-Up Access
One of the strongest insights from the discussion was that AI adoption cannot be mandated purely from leadership or emerge purely from grassroots experimentation. It needs both.
Hein described successful deployments as requiring a top-down and bottom-up model. Leadership must provide policy clarity, governance, and permission to use AI tools. At the same time, employees need broad access and simple ways to find immediate value.
“People need to feel safe that if they use AI in this way, that is supported,” Hein said. That is especially relevant in regulated sectors. Workers will avoid tools if they fear compliance violations, misuse of data, or unclear policy boundaries. Confidence drives adoption as much as functionality.
Compliance Is Becoming a Competitive Advantage
Many organizations still think of compliance as a blocker. Increasingly, it is becoming an enabler.
Hein emphasized Google’s long-term strategy of using accredited commercial cloud environments rather than isolating public sector customers in feature-limited siloed environments. That model is designed to give agencies access to frontier innovation while maintaining regulatory controls.
“If you need to have US data location, great—that’s an easy thing that we can configure,” Hein said.
This is an important shift. In the AI era, access to compute, GPUs, model ecosystems, and rapid innovation cycles matters. If compliance environments lag too far behind commercial environments, agencies risk falling permanently behind.
The more strategic model is compliant access to innovation, not isolation from it.
Open Model Access Matters for Government Flexibility
Another critical theme was optionality. Hein noted that agencies increasingly want access to multiple model types, including open-weight models and frontier commercial models, through a governed platform approach.
“Our overall ecosystem gives us access to all of the open weights models,” Hein said, referencing Google Cloud’s broader model access strategy.
That matters because no single model will fit every agency need. Some workloads require cost efficiency. Others require sovereignty, explainability, customization, or offline deployment to tactical edge environments. Public sector AI strategies built around one closed ecosystem could quickly become restrictive.
Legacy Modernization Is Becoming Continuous Improvement
Government environments often carry decades of technical debt, from mainframes to aging business systems. AI is beginning to change the economics of modernization.
Hein referenced agencies using AI-assisted development tools to modernize COBOL systems and rework aging platforms. More importantly, he framed modernization as an ongoing cycle rather than a one-time program.
“We can actually create continuous improvement using AI as a facilitator,” Hein said.
That mirrors what many enterprise leaders are discovering: modernization is no longer a project with an end date. It is a continuous operating discipline.
The Agentic Workforce Is the Next Big Story
When asked where things may be by next year, Hein pointed to the rise of AI agents as active participants in mission execution.
“We’re going to start to see this whole workforce transformation,” Hein said. “You start to see whole ways of how you accomplish your mission with these AI agents as a valuable part of the team.”
That framing is important. The future is not simply AI tools sitting beside employees. It is coordinated workflows where humans and software agents share responsibility.
For public sector organizations, that could mean faster permitting, improved case management, automated compliance processing, smarter citizen service operations, and accelerated internal administration.
Analyst Take
Public sector AI may be one of the most underestimated transformation stories in the market.
Governments face many of the same challenges as enterprises (i.e., legacy systems, talent shortages, fragmented workflows, and rising expectations) but often at larger scale and with stricter accountability. That pressure can become a catalyst when the right platforms emerge.
Three themes stand out. First, AI adoption succeeds when it starts with workforce productivity, not abstract transformation promises. Second, compliance-native innovation models will outperform isolated gov-only stacks that cannot keep pace with modern AI ecosystems. Third, the next stage of AI in government is not chatbots. It is agentic operations tied directly to mission outcomes.
If 2025 was the year of experimentation, 2026 and beyond may be the years agencies quietly become some of the most sophisticated AI operators in the market.
