In medical affairs, AI delivers its greatest value by changing how teams work, think, and measure impact for patients, rather than just speeding up existing tasks.
In a webinar hosted by Marcus Evans in collaboration with Envision Pharma Group, three industry leaders – Emma Vitalini, formerly of Amgen, Sarah Clark of Novo Nordisk, and Marie-Ange Noué of EMD Serono – joined moderator Nick Brown, VP of AI at Envision, to discuss practical and responsible applications of AI across scientific communications, insights generation, and operational workflows. The statements and opinions expressed in this content are those of the authors alone and do not represent the views, policies, or positions of their employers or affiliated institutions.
Throughout the discussion, the panel emphasized that effective AI must be grounded in data quality, governance, responsible scaling, and human expertise. When done right, it can elevate medical affairs from a reactive function to a predictive, insight-driven strategic partner.
The Q&A below captures direct perspectives from the panel, including early lessons learned, foundational challenges, compliance expectations, scaling realities, and the role of cross-functional partnerships.
What lessons have you learned from introducing AI into medical information and scientific communications?
Emma Vitalini: Start small and be very intentional as you roll these things out. Choose specific areas to automate or enhance, but ensure changes don’t disrupt downstream processes. For example, AI in call intake must integrate seamlessly with human follow-up. We realized that automating just one touchpoint – like intake – can affect the entire downstream response chain, so we’re careful to test end-to-end, not just in isolation.
Marie-Ange Noué: Don’t start with the tool – start with the intelligence gap you're trying to fill. Focus on measuring scientific influence, not just operational efficiency. AI should make us smarter and more strategic for our patients, not just faster.
Sarah Clark: Begin with user needs. Adoption depends on seamless workflow integration. And none of this works without quality inputs: garbage in, garbage out. We’re focused on making AI feel like a natural extension of the team – not a bolt-on. When people see the value without having to change how they work, that’s when you get traction.
Nick Brown: It’s also important to just get going on tangible projects. You can spend all of your time trying to find the perfect use case, but you are in danger of falling into analysis paralysis. It’s better to start and learn, build up new muscles, and discover more about the problem as you solve it.
What are the biggest workflow challenges in medical affairs before introducing AI?
Sarah Clark: AI amplifies what’s already there. If your scientific strategy or processes are weak, AI won’t fix them. Fix the basics first.
Marie-Ange Noué: Fragmentation of scientific knowledge is a major barrier. Data lives in silos – publications, customer relationship management, and safety systems. Without a unified framework, AI can’t connect the dots. We’re introducing the idea of return on intelligence (ROI) – not just time saved, but how AI accelerates decision-making. In many cases, the information we need is already in the system – just not in a way that AI can meaningfully use it. Structuring that knowledge is where the hard work begins.
Nick Brown: Business process mapping is essential. You need to understand your workflows before applying AI. This maturity helps you become data-driven in solving problems, and you can really quantify the return on investment at the end. This is critical for ongoing investments. Without that clarity, you end up investing in the wrong place or building tools nobody uses. It’s not about throwing AI at pain points. It’s about understanding the real friction, first.
How can medical affairs ensure AI remains compliant with evolving regulations?
Emma Vitalini: Engage early with legal, compliance, and data privacy teams. Establish both AI and data governance frameworks. Prioritize transparency, explainability, and traceability of sources.
Nick Brown: Build trust by involving stakeholders from the start. Use AI tools like knowledge graphs to reduce hallucinations and LIME to improve data transparency. The difficulties come when your AI solutions function as a black box. One of the best ways to prevent that is to use techniques that explain why the AI is making a decision – things like attention mapping, or curated knowledge graphs.
Sarah Clark: We need to understand how large language models (LLMs) are trained. With healthcare professionals now using these tools for education, transparency, and control over input sources is critical. It’s especially important in therapeutic areas where misinformation can spread quickly. If we don’t have confidence in how these models are trained, we risk introducing bias into medical education.
How do you articulate ROI and the value of AI to senior stakeholders?
Emma Vitalini: Efficiency gains are easy to measure, but AI also transforms how we work. Use proxies like hours saved, role avoidance, and improved decision-making, and partner with finance to quantify impact.
Marie-Ange Noué: Look at ROI – meaning that return on intelligence – and focus on strategic impact beyond efficiency. Consider how AI improves scientific positioning, accelerates decision-making, or changes external behavior. Senior leaders value foresight and influence more than automation.
Nick Brown: Start with impact, not just cost. We’ve seen AI reduce hours, improve response times, and streamline workflows, but what resonates most with leadership is how it changes the quality of decisions. That’s where we see true return on investment.
What are the challenges of scaling AI in medical affairs?
Sarah Clark: Change management is key. Co-creation and clear communication help facilitate the integration of AI into daily workflows. It’s not just the technology – it’s the human side.
Marie-Ange Noué: Pilots thrive on focus and tight governance. Scaling introduces complexity, including more stakeholders and compliance issues. You must scale the operating model, not just the technology. That includes resourcing, change management, and building the right governance as you grow.
Nick Brown: I don’t think many senior stakeholders fully appreciate that building the AI algorithm is just 5% of the work. The rest involves model management, infrastructure, monitoring, and governance. Most of the value comes later – when you’re monitoring performance, refining the model, and adjusting it as the business evolves. That’s the real operational burden of AI.
How can medical affairs teams collaborate effectively with AI partners?
Sarah Clark: Co-creation is essential, so seek transparent, nimble partners who evolve with you. Bring people along on the journey, communicate clearly what the impact will be, and never lose sight of the human element. The medical communications value chain is transforming, and vendors must not only adapt but also openly share how their workflows are evolving.
Nick Brown: There’s no one-size-fits-all. Whether building, buying, or borrowing AI, success depends on creative human experts and strong collaboration.
Marie-Ange Noué: One of the biggest mindset shifts is moving away from vendor management and toward co-creation. Our best outcomes come when internal and external teams collaborate closely from the start, with mutual understanding of the scientific and technical goals.
What emerging AI use cases show promise – and where should leaders be cautious?
Marie-Ange Noué: AI is evolving from an efficiency tool to a strategic partner. Predictive scientific intelligence and personalized engagement are high-value areas.
Emma Vitalini: Personalization at scale is exciting. But avoid over-relying on unvalidated external data. Always keep a human in the loop.
Sarah Clark: LLMs in healthcare education are promising but risky. Predictive diagnostics – if powered by integrated-health systems – could be a game-changer. We have to make sure we’re not creating unintentional bias or misinformation. Just because it sounds right doesn’t mean it is – and that’s dangerous in a medical context.
Nick Brown: Ultimately, a lot of exciting technologies are going to change the face of medical affairs in the near term, but what excites me the most is when a lot of the information will integrate – think of it like an operating system for medical affairs. Once we really have AI-enabled workflows, that’s going to really make a difference for patients. We’re not far off from workflows that are fully AI-assisted, from insight generation to content creation. When that happens, the whole medical affairs lifecycle becomes more intelligent and connected.
Where AI is delivering strategic value in medical affairs
Throughout the discussion, one message rang clear: AI is no longer just about efficiency, it is reshaping how teams deliver meaningful value to patients faster.
For medical affairs, success with AI means:
Starting with the right questions, not the right tools
Designing for explainability and trust from the outset
Embedding human expertise into AI workflows
Treating data quality and governance as strategic priorities for investment
Envision is proud to continue advancing the dialogue around responsible, scalable, and strategic AI to help the life sciences industry.
If you missed the session or want to revisit key moments, you can access the full replay here. To learn how Envision blends medical expertise with advanced AI models to elevate scientific strategy and impact, reach out to our team.