A physics-grounded pipeline for protein interaction.

Biology holds the answers to humanity’s hardest diseases, and the tools we use to interrogate it should be as precise as the systems they study. For the first time, we can model the physical forces that govern how molecules recognise one another, rather than guessing our way through them.

But today, discovering an antibody that binds where it should is still overwhelmingly a matter of trial and error. When we looked deeper into why, we found that current approaches, from screening to sequence-based prediction, treat protein interactions as patterns to be memorised rather than physics to be computed, to the point of being outright wasteful of time, capital, and biological material.

Actimo Labs exists to cure this waste, and to approach the precision of physics itself in predicting how proteins interact and eventually to design them from first principles.

Concretely, as of today, we have built two things:

• An energy-based representation that models protein-protein interactions through the physical forces that actually drive binding, rather than statistical correlations inferred from sequence alone.
• A target-aware design engine that proposes and ranks binders against a specific epitope, built to close the loop between prediction and wet-lab validation.

The reason for doing this is that in biology, sequence-based models scale with how much data they have seen, while physics scales with how well it captures reality. By grounding our predictions in energy and structure rather than pattern-matching, we generalise to novel targets where training data is thin or absent, exactly the frontier where most drug programmes stall.

Wet-lab friction is a major hurdle for computational drug discovery. It’s pointless if scientists have to run hundreds of blind experiments to find the one binder that works. That’s why, from day one, we’re building a lab-in-a-loop platform that compounds every experiment into validation data, so each campaign makes the next one sharper, a flywheel that turns wet-lab results into a widening moat.

We are a small team of biomedical engineers and computational drug design scientists with a tight focus: to bring this platform to life and build the tools around it, so therapeutics teams of any size can design the antibodies they need rather than stumble onto them.

We have early results where our engine designed and ranked binders against a defined epitope, as part of live fee-for-service work with therapeutics partners.