Phasing AI

Phasing is an applied research lab working on the intersection of AI and life sciences.

The next generation of medicines will be developed by organisations that are model-centric, built around robust data, and increasingly autonomous. Right now, the industry is held back by agonizingly slow processes, opaque data flows and insights trapped in one part of an organization. AI models are limited in their usefulness, lacking visibility into informal handoffs and a deeper understanding of scientific execution and management of long-running clinical projects.

We are working across the industry to change that. We study life sciences research and operations as they are actually practised, define what competent execution looks like, and build the settings in which AI systems can be measured against it. Right now, a model can read a protocol and reason about a deviation. It cannot yet carry the long chains of judgement a monitor requires, or anticipate and execute the number of small interventions that a project manager needs to manage all the different actors in a study. These skills cannot be taught from literature. They live in the records of past studies and experiments: the query that gets raised, the escalation that gets made, the deviation that turns out to matter. We start by capturing these learnings, where AI can learn as much from failures as from successes, and make the process more interpretable, robust, and scalable.

Our aim is not to replace human ingenuity, but to enable AI that is a better partner to humans in the process of scientific discovery and development.

We are a team of AI researchers, engineers, scientists, and operators. If you believe that rigorous, creative collaboration can solve the most meaningful problems in life sciences, we would like to talk.