What It Means to Become a Frontier Agency

Original broadcast 2/3/26

Presented by Microsoft

Carmen Krueger, Corporate Vice President, US Federal at Microsoft, describes “frontier agencies” as organizations that pursue productivity and innovation at the same time, rather than treating innovation as a separate effort or a future ambition. She explains that frontier agencies take advantage of modern AI capabilities while maintaining an intense focus on practical improvements—especially the kinds of measurable gains that can help agencies deliver services better, faster, and with greater accuracy. In her view, becoming a frontier agency is not simply about adopting the newest technology, but about using innovation in a disciplined way that directly strengthens mission performance.

Krueger emphasizes that frontier agencies approach innovation as an ongoing, daily reality. Instead of positioning innovation as a special project for a later date, they integrate it into routine work and accelerate the pace of change inside the organization. She argues that this shift is increasingly necessary because the speed of innovation today is accelerating far beyond what organizations faced even five years ago. This faster cycle requires agencies to remain agile and future-focused, building technology strategies that can evolve quickly and keep pace with emerging capabilities.

One important distinction Krueger makes is that private sector frontier firms often measure progress through market share, profitability, and competitive advantage, while federal agencies must assess progress through different outcomes. However, she notes that the idea of “competitive advantage” still applies in government, just in a different form. When federal agencies become more productive and strategically capable, the benefit flows outward to citizens, federal personnel, and warfighters who rely on government outcomes. She frames this as a national imperative: modernizing how government operates strengthens the country’s capacity to deliver at scale.

To define success, Carmen Krueger outlines several productivity-focused indicators agencies can track. She points to time savings for federal employees, measurable increases in accuracy, and faster cycle times for mission-critical work. These metrics can translate to improved delivery of benefits and services, as well as more consistent outcomes across large operational environments. She reinforces that frontier agencies must go beyond general enthusiasm for AI and establish real measures that show whether AI is helping people work better and whether mission outcomes are improving.

Krueger also identifies foundational behaviors that support frontier agency progress. She explains that effective AI adoption begins with strengthening the employee experience. AI should expand what employees can accomplish by increasing what she calls human decision-making power. She emphasizes that AI is not simply about replacing labor, but about enabling employees to make better decisions by giving them faster access to relevant information, providing assistance with routine tasks, and reducing the friction that slows down knowledge work. In this model, AI becomes a tool that augments human judgment and allows employees to contribute more directly to innovation.

Another major theme in the discussion is mission alignment. Krueger argues that innovation must be connected to mission outcomes or it will eventually collapse under its own weight. She describes the threat of “random acts of innovation,” where organizations chase new ideas or technology trends without clear mission relevance. In these cases, momentum breaks down because the organization stops trusting the value of the effort. She explains that disconnected innovation creates false positives, shiny-object distractions, and frustration among employees who do not see meaningful results.

Krueger returns to the role of business process modernization, emphasizing that many government processes were designed years ago under very different technological conditions. She notes that agencies will not achieve the value of AI if they attempt to overlay new AI tools onto old workflows without rethinking how the work should be done. Instead, she advocates for reshaping business processes to take advantage of the capabilities now available. This kind of modernization is essential to bending the innovation curve—creating faster cycle times, better outcomes, and productivity gains that agencies can sustain.

The conversation also highlights the importance of data in frontier agency success. Krueger explains that data democratization is important, but it must be implemented responsibly. She clarifies that democratizing data does not mean everyone has access to everything. Rather, it means agencies establish appropriate access controls and governance so that the right people can use the right data with confidence. She highlights the importance of data ontology in ensuring that users understand data context, origin, and meaning, because access without clarity leads to confusion, misinterpretation, or poor decision-making.

Krueger then offers concrete examples of outcomes agencies have already achieved through AI pilots. One of the most compelling examples she shares involves a large agency Copilot pilot that produced measurable improvements over a short period. That pilot delivered a 74% improvement in quality of work, a 75% productivity boost, and time savings of up to two hours per week for many employees. Krueger emphasizes that these results become especially significant when scaled across thousands of workers, creating meaningful mission value and freeing time for higher-impact tasks.

She also shares an example of AI accelerating complex document processing. In that case, an agency needed to process legal, contractual, and financial documents—work that is typically slow, labor-intensive, and difficult to scale. The original estimate for completing that effort was years, but the work was completed in weeks. Krueger frames this as an example of how AI changes the timeline of business processes and expands what agencies can achieve without adding staffing or shifting resources away from other priorities. She notes that because the individuals doing that processing have other responsibilities, reducing the time and burden of the effort has a compounding benefit.

A final example Krueger highlights involves application modernization, one of the persistent challenges across federal IT. She describes how an agency used GitHub Copilot to modernize the code of a legacy application. The developer involved cited a 25% to 50% reduction in coding time. Krueger points out that this type of coding automation is already common in major technology companies and argues that federal agencies should benefit from those same acceleration tools. She frames this as a practical pathway to modernizing government systems faster and making progress on long-standing modernization goals.

Across the conversation, the core message is that becoming a frontier agency is not a matter of experimentation alone. It requires productivity-focused outcomes, employee empowerment, mission alignment, process redesign, responsible data access, and a willingness to measure and scale what works. Krueger positions AI not as an abstract concept, but as a tool agencies can apply immediately to deliver measurable benefits and accelerate mission outcomes.