I recently attended the J.P. Morgan (JPM) Healthcare conference in San Francisco, which gave me an inside look at the key focus areas in 2025 (and beyond) for life sciences organizations.
I tend to view most (too many) things through the lens of technology. Agents are in the hearts and minds of nearly every person and every conversation for those that live in this business of putting the letters “A” and “I” together. But I did not hear that word (“Agent”) mentioned in any of the biopharmaceutical presentations delivered by CEOs (vendor presentations was a different story). To understand how technology, data science, and AI are actually being adopted, adapted, and creating impact, let’s first step back and broaden our view outside of technology to the corporate strategies that are driving organizations in healthcare.
The Looming Patent Cliff Is a DOOZY
$400 BILLION. That’s the number folks like to throw around for potential loss of revenue due to loss of exclusivity (LoE) in the next few years. It was top of mind at JPM as organizations (BMS, Pfizer, and Merck, for example) focused on the strength, diversity, and momentum of their late-stage development pipelines to counteract this. M&A also took center stage as companies make deep investments into new molecules from clinical biotechs and expand therapeutic areas like neurology (like J&J’s $14.6 billion dollar deal as an example) which, given the conditions we are beginning to see now that we are on the other side of COVID, is a no brainer (TERRIBLE pun for which I apologize).
Recharacterizing the Drug Development Lifecycle
In the context of the LoE, the mood was actually quite optimistic, with executives doubling down their focus on the nature of biopharma: discovery and development of novel therapies. Last year around this time, I touched on how the industry would adapt to a new speed of industry acceleration due to technology and AI shifts. This has been overwhelmingly evident in R&D. Clinical stage biotechs are now delivering on novel molecular candidates at five times the historical cycle of target discovery to clinical trials. Global pharma organizations are not just sitting back as the recipients of molecule acquisition, they are also accelerating and shifting their own internal pipeline.
It’s been incredibly exciting to work this year with some of the top companies on challenges with single-cell sequencing and spatial transcriptomics, predictive and generative chemistry in virtual screening, and ways they are changing how bench scientists develop and test hypotheses by asking questions to their molecular databases. So while the nature of the life science industry remains constant, the characterization of the drug lifecycle is starting to change as more organizations adopt and adapt digital technology.
Embracing a New Paradigm of Competitive Intensity in the Market
In contrast to how some companies are grappling with LoE that could impact up to 40%-50% of their current market revenue, others had quite a different strategy focus at JPM. Sanofi (with only around 2% of their current market hanging on the patent cliff), for example, highlighted the value of their real-world evidence data they are collecting in real time and how it builds both payer and physician confidence with their marketed brands, allowing them to continually focus on unmet need. There is “room for everyone” even under the new competitive intensity when we consider the heterogeneous and challenging biological penetrations and the fact that, in many critical therapeutic areas, less than 10% of patients receive the appropriate treatment.
Takeda’s CEO Christophe Weber also stressed this, specifically citing how they are developing AI algorithms and digital tools for diagnosis acceleration as part of their differentiated strategy. Companion diagnostics and approved AI tools for predicting drug indications is going to be a huge focus in 2025 and coming years, particularly with recent first-ever (draft) guidance on the use of AI in drug development in addition to existing guidelines of AI in medical device development.
The Inevitable Rise of AI Agents (and Change Management)
Sounds like the title of the next apocalyptic blockbuster hit, doesn’t it? If 2023 or 2024 is coined the “The Year of Generative AI” then 2025 most definitely is the “The Year of Agents.” The power and promise of agents is alluring. For example, some of the key priorities where I see agents making a play are to:
- Analyze and summarize vast biological datasets.
- Assess drug or molecule interactions and create hypotheses for new molecules.
- Assist with medical writing for clinical protocols in trial planning and design.
- Improve risk assessment, audit, and regulatory compliance across digital operations.
- Build queries of observational patient data sources to fuel real-world evidence studies and market launch success.
- Analyze market trends for targeted sales allocation and effort.
- Optimize content generation and personalization and create treatment pathway recommendations in marketing and market engagement.
It’s really easy to list out those and a hundred more (or ask a chatbot) in a few minutes, and it's even becoming simple to execute a PoC successfully. But AI agents aren’t just data science or technology tools, they are specifically aimed at changing last-mile consumption and how decision automation will change business practices.
Referring back to the new FDA draft guidance of AI to Support Regulatory Decision-Making for Drug and Biological Products, which lays out a seven-step Risk-Based Credibility Assessment Framework, guess what the shortest step with a total of two sentences and 57 words (out of a 23 page document) was?
Step 5: Execute the Plan. The word “credibility” was used 75 times. Credible AI was naturally the focus: ensuring the right context of use, the right utility, the right business question, the right monitoring framework, the right stakeholders. Execution is no longer the hard part and industry and technology organizations together need to consider how AI (predictive, generative, agentic, enter-new-buzz-word-here) need to work in the context of the business if we are going to see actual adoption this year.
Out of anything organizations focus on for 2025, change management and modernization of business processes (and if we don’t come to the agent design table with innate and deep understanding of the business processes we are trying to change) will determine if “The Inevitable Rise of AI Agents” is a horror movie or an uplifting feel-good family flick.
This research paper, “Empowering Biomedical Discovery With AI Agents” is one of the better articles I have seen, describing the concept of an “AI Scientist” and what the set of agents would look like to deliver this. They detail how an AI Scientist would be the most mature level of AI autonomy (over assistants and collaborators, heavily aligned to maturity in reasoning). As someone who spent the bulk of my early career in academic and consortia research, a common challenge is what we call the “research-to-practice” gap.
Do I think we will see AI Scientists this year? No. But I think if we remember to keep a focus more on the modernization of processes, navigating the necessary change management and the context of consumption vs. execution we will see this gap shrink rapidly in the next three to five years — right in time to usher in a new set of therapies to the market that represent a diversity of modalities that we have never seen before.