AI Redefining How We Discover Medicines
In this conversation, Dr. Siddhartha Mukherjee (Manas AI) and Yiannis Kiachopoulos (Causaly) examine how AI is being applied across this entire pipeline, not as a single model, but as a system of interconnected decisions.
.jpg)
Most drug programmes don’t fail in the clinic, they fail much earlier.
The root cause is often target selection: a decision made under uncertainty, across fragmented evidence from human genetics, preclinical models, and clinical data. When that decision is wrong, every downstream step, from pocket identification to molecule optimisation, compounds the error.
In this conversation, Dr. Siddhartha Mukherjee, Co-founder of Manas AI and Yiannis Kiachopoulos CEO and Co-founder of Causaly) examine how AI is being applied across this entire pipeline, not as a single model, but as a system of interconnected decisions.
They discuss:
- How target validity can be assessed by systematically aggregating and stress-testing evidence across modalities
- Why traditional search-based approaches break down in vast chemical space, and how generative, build-and-grow models shift the problem
- The limits of current AI systems in ADMET, where tacit medicinal chemistry knowledge remains difficult to formalise
- The challenge of orchestration: linking models together so outputs remain biologically and clinically coherent
- Why new therapeutic modalities, such as RNA-based constructs, are being driven by biological insight rather than algorithmic novelty
“The real word for AI in this domain is not artificial intelligence it’s augmented intelligence.”
A recurring theme is validation. At Manas AI, Causaly is used not just to inform early decisions, but to continuously sanity-check outputs, aligning predicted mechanisms with known biology, structural data, and clinical signals.
“If it’s real science, it can be augmented.”
This is a scientific perspective at where AI is already changing drug discovery and where the hardest problems remain.
Further reading
.png)
.png)
.png)
.png)
%20(14).png)
.png)
.png)
%20(5).png)
Get to know Causaly
What would you ask the team behind life sciences’ most advanced AI? Request a demo and get to know Causaly.
Request a demo