NVIDIA AI Podcast: How Agentic AI Shortens Drug Development and Boosts Patient Outcomes (Ep. 277)

Date: October 15, 2025
Host: Noah Kravitz
Guests:

• Raja Shankar – VP of Machine Learning, IQVIA
• Avanab Roy – VP of Commercial Analytics Solutions, IQVIA

Episode Overview

This episode explores how agentic AI is revolutionizing the pharmaceutical industry, particularly in accelerating drug development and improving patient outcomes. Expert guests from IQVIA, a global healthcare intelligence provider, discuss the adoption of intelligent AI-driven automation across the entire drug lifecycle—from R&D to commercialization—and share practical insights on tackling industry-wide challenges such as data silos, workflow inefficiencies, and patient impact. The discussion also highlights emerging partnership models, including IQVIA’s collaboration with NVIDIA, which aims to build scalable AI platforms for life sciences.

Key Topics & Insights

1. What Is Agentic AI and IQVIA’s Role

[01:24–03:53]

• IQVIA’s Mission:
  • Accelerate innovation in life sciences, facilitate drug access, and get crucial treatments to patients faster using data, technology, and analytics.
  • “IQVIA is the engine that is working with every life sciences company and healthcare organization to make that dream come true.” — Avanab Roy [01:46]
• Agentic AI Defined:
  • Agentic AI is distinct from traditional machine learning and generative AI. It not only understands and produces natural language but can perform complex, multi-stage tasks autonomously—using agents that can pass information and tasks among themselves.
  • “This allows us to build more complex systems with multiple agents where different agents do different tasks, and they talk to each other...” — Raja Shankar [03:53]

2. Evolution: Machine Learning → Generative AI → Agentic AI

[05:32–06:55]

• From Passive to Active:
  • Traditional AI: Recognizes patterns (e.g., image classification).
  • Generative AI: Creates content on demand (e.g., generate an image of a car).
  • Agentic AI: Executes actions (“Drive me to work”)—actively automating workflows.
  • “With agentic AI, now you’re asking, ‘Drive me to work,’ and the car takes you. That’s the next era.” — Avanab Roy [05:32]
• Agents accelerate and enrich previously manual, repetitive tasks such as launching products, understanding disease landscapes, and operationalizing commercial engagement.

3. Adoption Status & Barriers

[06:57–08:33]

• Pilot Phase to Scale:
  • The industry has had success with pilots but struggles with scaled, operational adoption due to both technological and organizational constraints.
  • “The pilots are a roaring success, but when it comes to large scale adoption, companies are struggling... It will take the whole village to bring the new digital persona into your home.” — Avanab Roy [07:07]
• Challenges: Change management, compliance, governance, and integrating digital "companions" into existing workflows.

4. Transforming Drug Development with Agentic AI

[08:39–12:40]

• Clinical Trial Simulation:
  • AI enables simulation of clinical development and trials before major investments, reducing risk and potentially increasing success rates.
• Workflow Automation:
  • Automating hundreds of manual, repetitive processes in clinical trials can dramatically reduce time to market and, consequently, provide financial and patient care benefits.
  • “If you're able to do that, we could reduce the time of clinical trials... The NPV extra NPV is in the hundreds of millions of dollars for them.” — Raja Shankar [09:38]
• Commercial Impact:
  • Compressing the time between drug launch and getting drugs to the right patients, by leveraging agentic AI to cut through data overload and operationalize outreach strategies.
  • “Too much data leads to chaos. People are trying to stitch this data together before they can track the insight, and your launch is stalled because of one decision...” — Avanab Roy [10:41]

5. Data: From Overwhelm to Insight

[12:40–17:43]

• Insight Gap:
  • The core challenge isn’t data scarcity, but connecting, accessing, and leveraging the right data for actionable insights.
  • “People are saying they are not missing data… the gap is connecting the data, bringing the data together for the right use case to drive the right insight.” — Avanab Roy [14:19]
• Enterprise Data Sovereignty:
  • Many valuable datasets exist within enterprises, not available to public models. There is a need for agents and domain-specific foundational models to unlock this potential.
  • “There is a lot of data that sits within enterprises... which dwarf the public data in some sense. The AI has not been trained on this data.” — Raja Shankar [15:12]

6. Impact on Patient Outcomes

[17:43–22:20]

• Optimizing Treatment:
  • AI and agentic systems can help personalize treatment plans, ensure better trial designs, decrease failure rates, and improve outcomes—provided data silos are broken.
  • “If they were able to pull this data together and run AI on this, we would design much better clinical trials than we are able to today... The technology exists today. The data exists today. The challenge is… organizational and silos.” — Raja Shankar [18:44]
• Continuous Engagement:
  • On the commercial side, agents support ongoing information flow to both physicians (HCPs) and patients, aiding adherence and education.
• Privacy Trade-off:
  • Reluctance to open access to national health datasets impedes progress, and current privacy/security vs. impact on lives trade-off needs better balancing.
  • “By not applying AI to these data sets today, it’s costing lives… we haven’t got that balance right today.” — Raja Shankar [21:13]

7. Advice for Companies Deploying Agentic AI

[22:20–27:16]

• Start with the Problem (Not Tech):
  • Identify core business challenges, define clear KPIs and outcomes, and don’t get “excited” about AI for its own sake.
  • “Start with a clear business problem. Don’t get excited because it’s a new AI. It’s like, ‘I have a hammer looking for a nail.’” — Avanab Roy [23:24]
• Fail Quickly & Iterate:
  • Quick pilots with defined decision gates, followed by tough calls on whether to continue.
• Data Readiness:
  • Ensure accessible and compliant data—don’t attempt massive data lake projects, but validate you have what’s needed for the pilot and beyond.
• Plan for Scale & Change:
  • Successful POCs are often technical; scaling requires cultural and organizational readiness, new skills, and adaptation.
• On KPIs and Benchmarking:
  • Historically, no clear benchmarks for “good” exist—agents may help create better standards for process quality.
  • “There is no gold standard of performance because… different people writing the same document will write it differently… With an agent, at least you can measure that performance.” — Raja Shankar [26:16]

8. The NVIDIA-IQVIA Partnership & The Future of Agentic AI in Healthcare

[27:16–31:50]

• Next-Gen Collaboration:
  • Both companies are “enablers” not competitors, uniquely positioned to create domain-specific, scalable agentic AI platforms for life sciences.
• The Data Flywheel:
  • Data, domain expertise, and tech combine to create a fast-learning “digital workforce” that rapidly improves and adapts.
  • “It’s like you have a graduate with all the theoretical knowledge, that’s the agent on day one. Then they go through 10 years of learning of the domain. But that agent can learn it in 10 days, right?” — Avanab Roy [28:06]
• Enabling Industry-Wide Acceleration:
  • Vision for catalyzing shared industry platforms, where life sciences companies benefit collectively without direct competition.

Notable Quotes & Moments

• On Agentic AI Evolution:
  • “With agentic AI, now you’re asking, ‘Drive me to work,’ and the car takes you. That’s the next era.” — Avanab Roy [05:32]
• On Data and Insight:
  • “There is a lot of data that sits within enterprises... which dwarf the public data in some sense. The AI has not been trained on this data.” — Raja Shankar [15:12]
• On ROI and Adoption:
  • “Start with a clear business problem. Don’t get excited because it’s a new AI... It’s like, ‘I have a hammer looking for a nail.’” — Avanab Roy [23:24]
• On Privacy and Patient Cost:
  • “By not applying AI to these data sets today, it’s costing lives.” — Raja Shankar [21:13]
• On Partnership Model:
  • “Our vision is to bring agents to life which are life sciences specific, who are trained on life sciences, understand what it takes to drive insights and mine data to serve the broader population or the life sciences workflows.” — Avanab Roy [28:06]

Timestamps for Key Segments

• What is Agentic AI / IQVIA’s Role: 01:24–03:53
• Agentic AI vs. Previous AI/ML eras: 05:32–06:55
• Industry Readiness & Barriers: 06:57–08:33
• Opportunities in R&D and Commercial Workflows: 08:39–12:40
• The 'Data Insight Gap' in Healthcare: 12:40–17:43
• Patient Outcomes and Data Silos: 17:43–22:20
• Advice for Adoption & Measuring ROI: 22:20–27:16
• The NVIDIA-IQVIA Partnership and the Future: 27:16–31:50

Summary and Takeaways

This episode demystifies how agentic AI stands to reduce drug development timelines and improve patient outcomes by enabling smarter automation and insight extraction across clinical and commercial domains. Success depends not simply on technology, but also on breaking silos, connecting data, setting clear goals, and reimagining industry partnerships. The NVIDIA-IQVIA collaboration signals a new era where scalable, domain-specific AI platforms can be built to serve—not compete with—pharma and healthcare organizations, with patient benefit as the central guiding principle.