Original Broadcast Date: 09/27/2026
Sponsored by Deloitte
The Health Resources and Services Administration is modernizing the way it collects and analyzes health information while preparing for a new phase of artificial intelligence adoption.
Srinivas Panguluri, acting chief information officer at HRSA, says the agency’s technology work spans both internal and external needs.
On the external side, data plays an important role in HRSA’s public health mission.
Panguluri says the agency historically collects and aggregates information, but aggregated data can limit what organizations are able to learn from it.
HRSA begins changing that approach several years ago by adopting the FHIR interoperability standard and building a platform around it.
The agency can collect information from medical encounters while anonymizing and de-identifying that information.
Panguluri says HRSA does not want to know who an individual is when the agency receives the data. At the same time, it needs information detailed enough to analyze trends over time and understand what is happening within individual health programs.
That longitudinal perspective can give the agency more actionable information.
HRSA works with health centers and technology providers as it builds the foundation. Pangaluri says the agency pilots applications and collects several million records to demonstrate that the underlying approach works.
The next step is applying that foundation to specific program needs.
Panguluri uses diabetes as an example.
Aggregated information might show a broad average, but it may not reveal where problems are emerging, which populations are seeing changes or whether trends are moving in a positive or negative direction.
More detailed information can help answer those questions, provided the data cannot be traced back to an individual.
HRSA is also building infrastructure to help analysts use that information.
Pangaluri says the agency has a modern data lakehouse where information can be stored in isolated and segregated environments. Analysts receive access only to the information they are authorized to use.
From there, they can build dashboards, conduct statistical analysis, perform time-series analysis and use other analytical approaches.
Panguluri says his philosophy as an IT provider is to give users flexibility rather than dictate a single analytical tool.
But that flexibility needs limits.
HRSA maintains a suite of available tools so users have choices while the agency retains enough control to secure and govern the environment.
Data governance is central to that model.
Panguluri says HRSA needs to understand where information originates, its lineage and its meaning. The agency uses tools including data catalogs and maintains a business glossary so analysts understand what the information they are using actually represents.
The same emphasis on governance is shaping HRSA’s approach to artificial intelligence.
Pangaluri says employees already use AI for lower-level tasks such as grammar correction and spreadsheet work. AI tools have been available for more than a year, and the agency tracks their usage.
The next stage is more ambitious.
HRSA is examining how AI can become part of workflows and reduce administrative tasks. That includes exploring agentic AI.
Panguluri says AI agents require a different governance approach because they can take actions rather than simply provide information.
He compares the issue to a research assistant. A human research assistant operates within understood behavioral limits, while an AI agent could potentially take unexpected actions.
That means agencies need to establish boundaries around what an agent is allowed to do.
Panguluri identifies several safeguards: clearly define those boundaries, create a “kill switch” and log activity so officials can understand what led to a decision.
Cost will also become increasingly important.
Panguluri notes that many AI tools are currently free or relatively inexpensive, but agencies need to watch token usage and other expenses as pricing changes and adoption grows.
Uncontrolled usage could produce unexpectedly large bills.
For Panguluri, that reflects a broader principle of managing government technology: assume something eventually will go wrong.
The important questions then become when it will happen and how much damage it can cause.
As HRSA expands its use of health data and AI, Panguluri’s approach combines experimentation with those guardrails—giving employees useful tools while maintaining control over data, security, cost and increasingly autonomous technology.