Last week at the AI in Clinical Development Summit 2025 in NYC, I had the pleasure of joining an...

Last week at the AI in Clinical Development Summit 2025 in NYC, I had the pleasure of joining an amazing panel with Sara Hughes (GSK), Rob Saiter (Accenture), and my former colleague Gabriel Eichler, moderated by Adam Petrich, MD from QuantHealth.

We tackled a timeless question: Build or buy?

Specifically - how should pharma approach AI innovation in clinical development?

Three takeaways from our discussion:

• Pharma has rich in-house data assets, but much of it remains siloed or unstructured, limiting its value for AI, if you are unwilling to invest in it; and external data sets are often purchased as one-offs for specific project and not leveraged strategically

• Internal development teams most often can't move at the pace of startups due to scale and complexity

• Off-the-shelf AI tools rarely succeed if they don't integrate seamlessly into clinical workflows

The consensus? A "build with" strategy, co-developing with selected strategic AI startup partners, strikes the right balance between speed, fit, and long-term value.

Thanks to everyone who joined us for the fantastic event, and to QuantHealth for hosting!