Lot of unknowns from the aspect of AI are around in life science. The product manager role is the most bullish role going forward.
Good product people understand that fundamentally as the cost to build goes down due to AI- they will focus to worry about a product discovery point of view.
Otherwise, if one goes to the costumers, quotes them and then use them to prioritize a product feature, it is product aggregation and it is a job to be easily automated by AI. That, in my opinion is no product management.
One thing that is clear- despite the market assumption of AI in life sciences or pharma is - AI needs to be democratized and that can only happen if we collectively leverage the energy
and the transparency of open source and open science. This will give everyone a
voice in what AI is, what it does, how it’s used, and how it impacts society and the scie ce we do today. It will ensure that the advancements in AI are not driven by the privileged few, but empowered by the many.
The future of AI is not just multimodal, but also multimodel. For too long the game of AI has been rigged with the game of scale- but the real great AI product lies at the intersection of size and efficiency. Therefore- there is a growing trend of fit-for-purpose genAI models with a tight data fabric associated to it.
I believe there is no physics law expecting AI to be expensive - so it is only an engineering problem to be solved.
You can argue all you want about new models being more benchmarked- etc (btw, benchmarking in itself is it's own whack a mole!). 20 years ago, we had a similar problem with the internet - with regards to trusting the search result of the engine. We see the same trend now. So, confirmation bias is a real thing- no matter where in the field you are.
Good product people understand that fundamentally as the cost to build goes down due to AI- they will focus to worry about a product discovery point of view.
Otherwise, if one goes to the costumers, quotes them and then use them to prioritize a product feature, it is product aggregation and it is a job to be easily automated by AI. That, in my opinion is no product management.
One thing that is clear- despite the market assumption of AI in life sciences or pharma is - AI needs to be democratized and that can only happen if we collectively leverage the energy
and the transparency of open source and open science. This will give everyone a
voice in what AI is, what it does, how it’s used, and how it impacts society and the scie ce we do today. It will ensure that the advancements in AI are not driven by the privileged few, but empowered by the many.
The future of AI is not just multimodal, but also multimodel. For too long the game of AI has been rigged with the game of scale- but the real great AI product lies at the intersection of size and efficiency. Therefore- there is a growing trend of fit-for-purpose genAI models with a tight data fabric associated to it.
I believe there is no physics law expecting AI to be expensive - so it is only an engineering problem to be solved.
You can argue all you want about new models being more benchmarked- etc (btw, benchmarking in itself is it's own whack a mole!). 20 years ago, we had a similar problem with the internet - with regards to trusting the search result of the engine. We see the same trend now. So, confirmation bias is a real thing- no matter where in the field you are.
What is a good AI product?
We are living in a moment where each organization is going from +AI to AI+ solution. Even MIDD has transformed from QSAR models plus PBPK models to AI-MIDD.
To create an innovative product which is a value creator is to first be a product that is an AI value consumer. And most of all any company which has worked through their data fabric - will be able to get their AI product produce viable value. When you think of data fabric, think information connectedness, self-service, ease of access and data protection.
A great example of an AI Value Creator is L’Oréal, one of the world’s leading beauty companies. Imagine the corpus of formulation, material science, and preference data L’Oréal has accumulated as it nears its 120th birthday. In essence, L’Oréal possesses data that defines the language of makeup. It wants to be an AI Value Creator, so it set out to create a private AI model (in collaboration with IBM) to accelerate tasks like the formulation of new products, reformulation of existing cosmetics, and
optimizations to scale-up production. If L’Oréal was just an AI User, it would give this data away, but instead it views its data as a competitive advantage and decided to put it to work to better equip L’Oréal’s 4,000 researchers worldwide over the next several
years. L’Oréal isn’t just applying AI to beauty—it’s giving it a makeover of its own.
The Product Manager’s True Value: Product Sense
A product manager’s worth lies in their judgment—their ability to see the world not as it is, but as it could be. This product sense is as rare in PMs as creative problem-solving is in R&D scientists. The opportunities for AI to transform life sciences are endless, but it’s important that you make all your efforts about business opportunities and outcomes, and not data science projects.
There are three kinds of product managers:
- Delivery PMs: Scoped to chase features and check boxes. Necessary, but limited.
- Feature Team PMs: Stewards of product features, pruning or growing to avoid bloat.
- Product Discovery PMs: The visionaries. They don’t just understand customers. They know the business, the science, and the tech. They ask: How does AI reshape our product line? Is our strategy still sound? How does it impact our people?
The best product leaders don’t just react to enabling tech—they anticipate its ripples. They run hackathons to upskill teams, hire for vision, and ensure their people can execute. They know the future isn’t just about AI—it’s about data as a product, treated like an API. Every business stakeholder, from scientists to executives, becomes a domain expert responsible for high-quality, accessible data.
The Human Element in AI Products
What’s changing fastest is the rise of human features in AI products—elements of user experience that were never considered before. UX design is splitting into two paths: agent-centric (where AI drives the interaction) and user-centric (where humans remain in control). This shift demands product managers who can balance technology’s potential with human needs.
So what makes a successful AI product?
AI SUCCESS = MODELS + DATA + GOVERNANCE + USE CASES
