Photo courtesy of Sairohith Thummarakoti
Sairohith Thummarakoti, industry advisory board member at Texas A&M University–Kingsville, tells MobiHealthNews about his upcoming session at the 2026 HIMSS Global Health Conference & Exhibition in March, where he will present a framework for deploying agentic AI as a safe and effective teammate in clinical care.
MobiHealthNews: Can you tell us a bit about the blueprint you'll present and what parts of clinical care you will focus on?
Sairohith Thummarakoti: In this talk, I will share a blueprint for using agentic AI and low-code platforms to make claims processing and prior authorization behave more like a self-driving system and less like a manual back office.
The architecture is event driven across eligibility, benefits, claims, clinical documentation and payment systems, with a provenance graph that records every step an agent takes: what data it looked at, which rules and policies it applied, what it decided and why.
Agents run on top of a low-code workflow and case management layer and operate under signed contracts that define their scope, including which claim segments they are allowed to touch, what dollar limits they can auto-approve or -deny, what PHI they can see, and what cost and equity budgets they must respect.
I will focus on the end-to-end claims journey, including intake and enrichment, clinical and policy validation, straight through adjudication, prior authorization, fraud or waste or abuse signals, and appeals. The idea is to let low-code, agentic workflows handle the bulk of routine, low-risk volume without human intervention, while automatically routing only ambiguous or high-impact cases to human reviewers. Attendees will see how you can move toward no touch claims where it is safe but keep humans exactly where they add the most value.
MHN: Agentic AI are autonomous AI systems that independently set goals, reason, plan and take actions to achieve a specific objective without human oversight. Why do you think agentic AI is ready for use in healthcare?
Thummarakoti: Agentic AI is especially ready for the administrative side of healthcare, including claims, prior authorization and utilization management, because these workflows are policy driven, heavily documented and already tracked in terms of turnaround time, accuracy and compliance.
In the last few years, we have gained important building blocks such as reliable retrieval over benefit designs and medical necessity criteria, fine-grained access control and evaluation frameworks that measure error rates, bias and cost per transaction. Low-code platforms now make it possible to combine these capabilities with event-driven workflows and case management in a way that is repeatable and easier to govern.
This makes it possible to give agents very specific goals, such as fully processing all low complexity claims under a certain dollar amount within a defined time, inside strict guardrails. Agents do not invent new rules. They read your benefits, policies and clinical guidelines through governed connectors on a low-code platform, apply them consistently, show their evidence, and either finalize the claim or escalate it.
In other words, we are not talking about unconstrained autonomy over bedside care. We are talking about targeted autonomy over repetitive, rules-heavy claim decisions where performance can be measured and systems can be rolled back safely if something drifts.
MHN: What do you hope attendees learn from your talk?
Thummarakoti: I hope attendees leave with a very concrete picture of what agentic AI for claims looks like in production, and how a low-code platform can help them get there faster without losing control. That includes a reference architecture with an event mesh, a lakehouse with a vector index for policies and clinical content, agent contracts that define scope, a provenance graph for full traceability and an AI gateway that governs all model and tool calls.
On top of that, we will look at patterns for straight through processing, intelligent routing and automated appeals preparation, all with clear audit trails and equity checks.
Just as important, I want them to see a phased path from where many organizations are today, with manual queues, brittle rules engines and isolated AI pilots, to a future where most clean claims can be processed end-to-end with little or no human touch.
Low-code, agentic workflows can take on the routine work, while humans focus on complex clinical judgment, member experience and continuous improvement. If attendees walk away thinking that they can do this safely and can name the first low-risk claims flow they would pilot with an agentic, low-code architecture, then the session has achieved its goal.
Sairohith Thummarakoti's session "Reimagining Healthcare Infrastructure with Agentic AI and Cloud Computing: Toward Scalable, Ethical, and Personalized Systems" is scheduled for Wednesday, March 11, from 9:45-10:45 a.m. in Palazzo J Level 5 at the Venetian at HIMSS26 in Las Vegas.
