Transcript
Winter Circle Interview: Transforming Healthcare Operations with AI
Speakers:
- Russ Fordyce (Host)
- Madan Moudgal (Sagility)
[00:00–00:24] Introduction
Russ: Welcome back to the Winter Circle, everybody. I’m here today with Madan from Sagility. Today’s conversation is all about healthcare and how healthcare operations are being transformed with AI. Madan, welcome — and congratulations again on the AI Excellence Award.
Madan: Thank you, Russ. I’m pleased to be here. We’re very excited about the recent win for our Nurse Assist solution, and I’m looking forward to sharing more information about it with you.
[00:24–01:17] Sagility Overview
Madan: Sagility is a healthcare operations company. We provide services exclusively in the United States, supporting healthcare organizations such as health insurers — commonly referred to as payers — as well as health systems and providers. We are a technology‑led business operations and services company and have been operating in this space for about two decades.
Over that time, we’ve built deep expertise in the U.S. healthcare industry and its complexities. We’ve combined that domain knowledge with technology to develop a range of solutions as part of our long‑term transformation journey.
[01:17–02:20] Why Healthcare Is Hard to Modernize
Russ: Healthcare — and especially the payment systems in the U.S. — is extremely complex. It’s often viewed as one of the hardest industries to modernize and slow to adopt technology. Why is that, particularly when it comes to AI?
Madan: That’s a fair observation. I’ve spent nearly my entire professional career — almost 40 years — in healthcare technology. While the industry historically lagged in tech adoption, over the last five years it has made significant progress.
However, healthcare is highly regulated and very sensitive, particularly around patient data privacy. Those regulatory requirements make technology rollouts more challenging than in less‑regulated industries.
[02:20–03:32] Legacy Systems
Madan: One of the biggest challenges to transformation is the presence of large legacy administrative systems. These core platforms were often built over decades, and replacing them is extremely difficult.
Because of this, most transformation efforts must work within the constraints of existing infrastructure. Any new technology has to account for — and operate alongside — these legacy systems rather than fully replacing them.
[03:32–05:00] Prior Authorization Challenges
Russ: One of those legacy processes is prior authorization. Why is it such a significant challenge in healthcare?
Madan: Prior authorization is a sensitive issue. It exists largely because the industry faces substantial fraud, waste, and abuse, which is well documented.
The challenge is finding the right balance — containing costs and managing utilization without restricting or delaying a patient’s access to necessary care. It’s a difficult trade‑off with major cost implications for the healthcare system as a whole.
[05:00–06:06] Augmented Intelligence vs. Automation
Russ: You use the term “augmented intelligence” instead of “automation.” Why is that distinction important?
Madan: Simply put, it’s about keeping humans in the loop. We don’t believe these processes should be fully automated. Critical decisions related to care authorization must ultimately be made by people.
AI’s role is to augment that decision‑making process — making it faster, more accurate, and more efficient — rather than replacing human judgment.
[06:06–08:04] Defining AI’s Role
Madan: Determining what AI can do versus what requires human involvement depends on deep domain knowledge, regulatory guidance, and process complexity. Regulations at both the state and federal levels define what can and cannot be automated.
Simple, low‑risk interactions are good candidates for automation. More complex cases — particularly those affecting access to care — require human involvement, with AI serving as a decision‑support tool.
[08:04–09:22] Nurse Assist Overview
Russ: Give us an overview of Nurse Assist. What does it do?
Madan: Nurse Assist was developed in collaboration with our clinical experts and registered nurses who support major payers and providers across the U.S.
The solution extracts and summarizes relevant information from medical records — such as diagnoses, treatments, and pre‑existing conditions — and compares that information to clinical guidelines. It then provides a recommendation, while the registered nurse makes the final decision.
[09:22–10:11] Domain‑Specific AI and Trust
Russ: You’ve chosen a domain‑specific AI model rather than general‑purpose AI. Why?
Madan: Trust and relevance are critical in healthcare. While general‑purpose models are improving, healthcare requires specialization. Clinical language models designed for specific use cases provide greater accuracy, consistency, and reliability, which builds trust with clinicians.
[10:11–11:00] Safeguards and Compliance
Madan: We use strong guardrails to protect personal health information and ensure compliance with healthcare regulations. AI is restricted to approved workflows and permitted actions only.
These safeguards help ensure that innovation does not compromise privacy, safety, or regulatory compliance.
[11:00–13:26] Impact on Review Time and Experience
Russ: You’ve significantly reduced review times. What does that mean in practical terms?
Madan: Traditionally, nurses may spend 30 to 60 minutes reviewing lengthy medical records — sometimes hundreds of pages long.
Nurse Assist summarizes and highlights relevant information and directs reviewers to what matters most. In some cases, this reduces average review time by up to 50 percent, improving speed, efficiency, and overall patient experience.
[13:26–15:27] Integrating with Legacy Environments
Madan: Successful AI integration depends on three main elements: data curation, AI governance, and change management.
Data from legacy systems must be contextualized so AI can properly interpret it. Governance ensures compliance with organizational and regulatory policies. Change management helps teams adapt workflows, which is essential for successful transformation.
[15:27–17:08] Managing Risk in Healthcare AI
Russ: How do you balance moving quickly with managing risk?
Madan: You can’t rush healthcare innovation. Fundamental software development practices — especially testing — remain critical, even with generative AI.
Incremental rollouts, proof‑of‑concept testing, and validation are essential in a high‑stakes healthcare environment where accuracy and trust are paramount.
[17:08–18:40] The Future of Automation
Russ: Will healthcare ever be fully automated?
Madan: I believe healthcare will remain augmented rather than fully automated. The industry is complex and highly regulated, and human judgment will always be essential.
The most effective approach is selectively automating parts of workflows while keeping humans involved where decision‑making matters most.
[18:40–28:25] Other Areas Ready for AI
Madan: Beyond clinical review, AI is also transforming administrative interactions, member engagement, claims processing, and payment integrity.
Consumers are increasingly comfortable interacting with virtual agents as long as the information they receive is accurate and timely. Over time, this reduces cost, waste, and friction across the healthcare system.
[28:25–28:47] Closing
Russ: Healthcare may not become fully automated, but it’s clearly transforming — and quickly.
Madan: Absolutely. Even the next five years will bring significant change.
Russ: Congratulations again on the award.
Madan: Thank you, Russ. I enjoyed the conversation.
