Federal agencies should prioritize interoperable platforms, strong governance, and more disciplined use of data as they modernize systems with artificial intelligence (AI), according to a senior federal AI expert during a recent GovCIO webinar.
Anil Chaudhry, a senior adviser on AI for the U.S. government, explained that agencies are moving beyond the traditional “people, process, technology” model toward a “people, process, platform” approach that better supports long-term adaptability.
He pointed to the administration’s updated Presidential Management Agenda as an important reference for understanding federal management and governance priorities, noting that technology decisions are closely tied to how agencies oversee systems and operations.
“For a long time, we talked about people, process, technology,” Chaudhry said. “Now we actually are in the people, process, platform business [because] picking the right platforms is more critical now than it is picking the right technologies.”
Chaudhry explained that while many technologies can work across different platforms, the platforms themselves are not always interoperable. Poor platform choices, he said, can limit an agency’s ability to adjust to changing requirements.
Chaudhry also stressed the importance of modular, open architectures to reduce the risk of vendor lock-in. While licensing costs may appear low, he said, integration and migration costs can make it difficult to move away from platforms that no longer meet agency needs.
He also addressed the challenge of monitoring complex systems, saying agencies need domain expertise to define what normal operations look like so AI models can identify meaningful anomalies without overwhelming staff with alerts.
He also outlined an analytics maturity model that begins with descriptive analytics and progresses through diagnostic and predictive stages before reaching prescriptive analytics, where AI systems can recommend actions.
“The final … is the prescriptive,” Chaudhry said, cautioning agencies against collecting and processing data without a clear purpose, saying not all data adds value.
He emphasized the need for tools that help organizations identify which data supports their business objectives before investing in analytics capabilities.
Overall, he said consistent data platforms can provide a common operating picture across systems, allowing humans to respond more quickly and consistently. Specialized tools and AI models can then be built on top of those platforms to address specific mission use cases.
“It’s the power of the platform that really allows humans to act in a consistent way and to also be able to react faster,” Chaudhry said.