The pharmaceutical industry has largely moved beyond the ‘why AI’ question. The discussion is no longer about ‘whether’ teams should adopt artificial intelligence as they prepare for launch, but ‘how’ AI applications can be implemented and what launch challenges they’re best positioned to address. This was the conversation at the recent webinar Beyond the Buzz: Practical AI for Launch Impact, moderated by Impatient Health’s Gareth Morell and featuring Pfizer’s Prem Sundivakkam alongside Within3’s Lance Hill and Jason Smith.
The session was structured around three critical areas – each featuring a live poll designed to ground the conversation in real audience experiences.
- Multimodal Data: unifying diverse, fast-moving medical insight streams
- Impact Analysis: shifting from activity reporting to predictive, actionable impact
- Communication & Collaboration: routing insights effectively across the organization
For those who missed the webinar, we’ve summarized the session here and pulled out some key takeaways that demonstrate how medical affairs teams can use AI for earlier alignment, stronger insight generation, and more confident launch decisions.
Takeaway #1: Solving the challenge of multimodal data integration
“There are so many different data types and structures – all collected by different people for different purposes.”
– Gareth Morell, Impatient Health
As you prepare to launch a new therapy, there are numerous questions to answer. You might have a business problem that needs solving, or an unmet patient need to address. You might be looking for a new indication for a particular molecule, to understand what patients are saying, or to solve any combination of these questions (plus countless others). The answers are found across a wide variety of information sources – different data streams and datasets. Multimodal data integration is the process of harnessing all these diverse information streams simultaneously, allowing you to address critical launch questions holistically.
After setting the scene, the Beyond the Buzz panel discussed why multimodal data integration is so important. In Lance’s memorable words, teams are “in the area of the unknown” at launch and seeking greater understanding. 24 months down the line, they need to be in a position where they can create an education plan capable of ensuring all stakeholders understand how a new asset will help their patients. An inability to integrate and synthesize data from multiple sources means an inability to understand the market.
“If you start off and you misread that part of the market, all that work you’re going to do – all that heavy lifting – is going to be off target.”
– Lance Hill, CEO & Founder, Within3
But as the panel points out, markets aren’t static, and so the process of market understanding will never be set-and-forget. Prem noted how AI provides a constant stream of insights across different levels and at different points in time – allowing teams not only to act early, but to continuously adjust their launch strategies as market forces dictate.
“It’s actually helping us to detect contextual shifts in real time, and route these validated, equity-aware insights to the right owner. In that way, we can avoid missteps and demonstrate the impact on beliefs, behaviors, and patient outcomes.”
– Prem Sundivakkam, Executive Lead, Global Medical Communications and Congress, Pfizer
Takeaway #2: Accelerating the insight-to-action cycle
At this point in the discussion, the panel turned to how teams can accelerate the insight-to-action cycle—a clear priority for the audience. Poll results showed that 47% of respondents identified a faster insight-to-action cycle as the most important outcome of achieving multimodal data integration. Jason Smith underscored a critical nuance: not all data sources are created equal. Turning insight into action requires placing each data set in context and aligning it to a clear strategic narrative, rather than treating all inputs as equally actionable.
This is one of the major roadblocks that slows the insight-to-action cycle, but it’s also an area where AI excels. “You can look across hundreds of thousands, millions, if not billions of data elements in minutes,” Jason said. “That speed is something we’ve just not had before.” But for launch teams, it’s not as simple as adopting a large language model (LLM) like ChatGPT or Claude and setting it loose on the data. “They’re great at summarization, but they’re looking at the corpus of the internet, essentially,” Jason explains. Instead, pharmaceutical teams need the specificity of so-called small language models (SLMs).
“Understanding how a physician is speaking about a patient’s reaction to a drug, or understanding how patients are feeling about a drug – those are very nuanced. LLMs just aren’t contextualized. They’re very broad. They don’t understand the specialization of the language and terminology we use day in and day out.”
– Jason Smith, CTO, AI & Analytics, Within3
The webinar attendees wanted to know whether pharma organizations are ready to adopt SLMs like the ones the panel discussed. Jason sees this question as a double-edged sword for pharma organizations. While AI literacy is increasing in pharma and beyond, there are top-down pressures to deliver transformative impact right away. Stakeholders want to see an immediate return on their investments, and as much as the onus is on internal teams to demonstrate the value of the AI models they’re adopting, it’s also on the technology companies themselves to deliver on that initial promise. Jason describes it as “making these tools easier to align” with pre-existing technologies and strategies. Pharma organizations and the tools themselves are coming towards one another: ‘readiness’ is where AI literacy and ease-of-use meet.
Takeaway #3: What medical affairs teams should be measuring
“Measuring value in medical affairs is notoriously difficult – it’s probably notoriously ignored. You might say the perfect answers aren’t there, so we’re not as interested.”
– Gareth Morell, Impatient Health
The final point for the panel to discuss was the thorny issue of value metrics in medical affairs. Gareth summarized the issue very clearly with the above quote, but Perm believes the problem stems from a lack of a unified strategic narrative – and as a result of poor cross-functional collaboration. “We need cross-functional collaboration and to align on the narratives,” he explained, “because the impact measurement fails if medical can’t create a unified narrative with cross-functional teams. We have to clearly align on that. One of the things I often see is a lack of shared strategic context – we all need to align on one context, one goal, one objective.”
Lance explored the difference between metrics that are comparatively straightforward to capture – ‘activities’ like HCP education – and things like clinical behavior and patient outcomes, which are less tangible but arguably more important in determining launch success. AI can help to build a bridge from one to another – looking beyond the raw data (e.g. 73 KOLs educated in the past month) to more valuable insights with real strategic impact (the true patient reach of each of those HCPs, the behavioral impact those educational activities have had, etc).
“Where AI is really effective here is not just looking over time, but looking quickly at patterns and finding those signals much earlier.”
– Jason Smith, CTO, AI & Analytics, Within3
Conclusion: Putting insights to use
The final phase in implementing AI for launch impact is ironically where a lot of organizations traditionally fall down. They’ve gathered some valuable insights and meaningful outcomes data, but those insights ultimately go nowhere. For whatever reason, they aren’t put to use. “You have to understand that what you’re doing is telling a story around how the market’s behaving, and you’re telling that story to different audiences,” Lance explains. The disconnect comes when medical, commercial, market access and other departments are reading different pages, or different stories entirely. Inevitably, critical insights don’t make it to the right stakeholders because they don’t seem relevant to the story the individual department is telling. There’s no overarching strategy guiding the actions of each department in unison.
“The most effective organizations are those that have some sort of cross-functional group that’s on the same page with strategy, and can translate that strategy down into the different groups”
– Lance Hill, CEO, & Founder, Within3
Lance went on to explain how medical and commercial teams think differently from one another. Medical teams are very data-oriented; data points lead to conclusions. Commercial, on the other hand, starts with the “so what” and seeks to prove or disprove that question with data. These groups consume information in a different way, and they inevitably respond to insights differently, too. “Insights reporting is about getting the right information to the right audience,” says Lance. “Tools can help you do that, but ultimately it’s a communications issue more than it is a data or a science issue.”
The Within3 Launch Intelligence™ Platform provides a real-time market view capable of delivering the critical insights you need to achieve launch excellence and adapt your launch strategy at speed. Better yet, the platform is supported by your own dedicated client success team, ensuring you not only access the insights you need, but understand them – and put them to use. Want to know more? Schedule a demo today.
