Why Most Pharma AI Will Fail Without This One Thing
Most pharma companies are racing to apply AI across drug discovery, development and commercialisation, but many of those efforts will fail for one simple reason: the data underneath is not good enough. In this episode, Dr Andree Bates speaks with Lisa Downey, CEO of DrugBank, about why trusted, structured biomedical intelligence is the foundation pharma AI cannot succeed without.
Lisa explains how DrugBank has spent 20 years building and continuously curating a biomedical knowledge layer across drugs, targets, diseases and trials. With more than 156 million structured data points and over 60,000 academic citations, DrugBank is not just another dataset. It is a continuously maintained reference system designed so AI can reason over biomedical knowledge with traceability and trust.
The conversation explores why most pharma AI projects fall short. Lisa argues the blocker is rarely the model. Instead, teams hit the wall because internal data lakes are not harmonised, licensed third-party data may not be AI-ready, and public data sources are incomplete or not maintained for enterprise use. Brilliant ML teams then spend most of their time cleaning and reconciling data instead of creating real scientific or commercial value.
Lisa also breaks down what pharma buyers should test before trusting any AI vendor: interoperability, harmonisation, evidence lineage and continuous validation. She explains why human pharmaceutical expertise still matters, introducing DrugBank’s “human over the loop” approach, where experts set scientific boundaries, validation criteria and judgement so AI can scale inside trusted guardrails.
Topics Covered
Why most pharma AI projects fail before they scale
Data quality as the foundation of trustworthy AI
DrugBank’s 20 years of curated biomedical intelligence
Internal data lakes, third-party data and public data limitations
Why hallucinations often start upstream of the model
How to evaluate data quality: interoperability, harmonisation, lineage and validation
Human over the loop vs human in the loop
Why defensible AI needs traceable sourced facts
The difference between confident AI and grounded AI
Why proprietary context matters more than raw data
Eularis helps pharma and biotech leaders turn AI activity into board-defensible strategy and measurable commercial outcomes.If your organisation has plenty of AI in motion but very little that moves the commercial needle in a way the board can see, start with our 10-Day AI Diagnostic Sprint. It’s a focused diagnostic that surfaces what’s actually broken and what’s blocking results, before you invest in a larger strategy effort.
The Sprint diagnoses the problem. The AI Strategic Blueprint that follows is where we build the board-defensible strategy and plan.Details at eularis.com.
AI platforms and tools solve specific problems. Strategy makes sure you’re solving the right ones, in the right order. If you want help mapping priorities as you evaluate what to roll out next, send me a LinkedIn DM starting with ‘PRIORITIES’ and two lines: what’s already in flight, and the decision you’re trying to make next.
About the Podcast
AI For Pharma Growth is the podcast from pioneering Pharma Artificial Intelligence entrepreneur Dr Andree Bates, created to help pharma, biotech and healthcare organisations understand how AI-based technologies can save time, grow brands, and improve company results.This show blends deep sector experience with practical conversations that demystify AI for biopharma leaders, from start-up biotech right through to Big Pharma. Each episode features experts building AI-powered tools that are driving real-world results across discovery, R&D, clinical trials, medical affairs, market access, regulatory, insights, sales, marketing, and more.