SandboxAQ Unveils AI Tool to Accelerate Drug Discovery

The tool, now generally available via Claude’s Model Context Protocol, functions by ranking molecules against disease targets at a cost of roughly $1 per 1,000 comparisons. Unlike traditional computational methods that provide a single, often opaque score, AQPotency includes confidence intervals. This allows researchers to distinguish between high-certainty predictions and areas where the model’s reliability is lower, turning raw data into actionable decision-making for biopharma pipelines.

Beyond screening, the model operates in reverse. By inputting a single molecule, scientists can scan a panel of proteins to identify potential targets, effectively reverse-engineering the mechanism of action for unknown compounds. This capability has already seen practical application, with the model deployed in eight customer programs featuring experimentally validated results. Academic partners, including researchers at the University of Dundee and Columbia University, have leveraged the technology to navigate complex biochemical spaces, specifically targeting difficult membrane proteins linked to Parkinson’s disease. As SandboxAQ scales the platform, further integration with Google Cloud’s Marketplace is expected to expand access for discovery teams worldwide.