In his appearance on Fed Gov Today with Francis Rose, Terry Halvorsen, Vice President for Federal Client Development at IBM and former Chief Information Officer at the Department of Defense, offers a clear and practical look at how artificial intelligence is helping federal agencies regain momentum after the government shutdown. Halvorsen discusses AI’s real value today, the data challenges agencies must confront, and the compressed timeline federal leaders now face as they push to get operations back to normal.
Halvorsen begins with a straightforward observation: agencies that already have AI tools deployed hold a significant advantage as operations resume. AI allows employees to “look at their data much faster,” he says, and that kind of speed matters when staff return to large backlogs and tight deadlines. While Halvorsen is careful to emphasize that “AI is not replacing people,” he explains that it helps people do more work faster—reviewing records, surfacing insights, checking data quality, and supporting quicker analysis.
He sees this as the biggest advantage for agencies coming out of a shutdown. Decisions that once took weeks can be made far more quickly, and employees can move through backlogged tasks with greater efficiency. But he also stresses that this advantage only appears when data is ready for AI to use.
That’s where he identifies one of the federal government’s major challenges. Even though government data is often “richer and more structured” than corporate data, Halvorsen says it is not always AI-ready. He explains that the issue is not the amount of data—federal agencies have more than enough—but the structure and quality of it. Government organizations often feed AI overly large datasets instead of carefully curated ones, assuming bigger will be better. “It’ll get you an answer,” he says, “but maybe not as specific or as fast as you want.”
Halvorsen advises agencies to focus on curating smaller, cleaner datasets, ensuring accuracy, and structuring information properly. Doing so allows AI tools to perform well, deliver specific insights, and produce reliable outcomes. He believes that as agencies face the pressure of clearing backlogs and regaining operational rhythm, the temptation will be to rely on whatever data is immediately available. But he urges leaders to resist shortcuts and instead invest time into preparing their data the right way.
Another factor Halvorsen highlights is the calendar challenge federal agencies face coming out of the shutdown. He notes that the government is only about two weeks from Thanksgiving and around six weeks from the end of the calendar year—a period when many employees take extended leave. Even with AI, agencies have limited time to recover before holiday staffing levels drop. Halvorsen says this compressed timeline “is going to complicate this more than it would have” under normal circumstances, simply because key personnel may not be available consistently in the coming weeks.
Despite the pressure to get short-term tasks done, Halvorsen encourages leaders to think beyond the immediate scramble. He acknowledges that after a shutdown, it is easy to fall into the mindset of “just keep the trains running on time.” But he argues that agencies should also use this moment to set up long-term productivity gains. That means preparing data, strengthening analytic tools, and building a foundation that supports sustained AI use, not just quick fixes.
Halvorsen suggests agencies divide their goals into two time horizons: six months and one year. Some solutions should be achievable within six months, especially those focused on improving current processes or clearing backlogs. Longer-term plans—such as deeper modernization efforts or major data architecture improvements—will likely take a year or more. By dividing efforts in this way, agencies can make sure immediate needs are addressed without losing sight of larger transformation goals.
He points out that many federal leaders have already been preparing for this moment. Even during the shutdown, some agencies continued planning and evaluating the data sources they would prioritize once operations resumed. Halvorsen sees this as critical: the faster agencies start organizing their datasets and running analytic tools, the faster AI will begin producing meaningful value.
Throughout the conversation, Halvorsen maintains a balanced and pragmatic tone. He understands the urgency federal agencies feel, yet he continually brings the discussion back to the importance of data quality, long-term thinking, and smart preparation. AI can accelerate recovery, he says, but it cannot compensate for unstructured information or poor planning.
In the end, Halvorsen frames AI as both an immediate tool for recovery and a long-term catalyst for modernization. By combining thoughtful data management with realistic timelines and a focus on both short-term and long-term outcomes, agencies can turn what could be a difficult period into a chance to build a stronger, more responsive, and more data-driven government.