Artificial Intelligence Management Systems
ISO 42001 is the international standard for AI management systems. Here's how its principles guide our use of proprietary AI models within our molecular design service.
What is ISO 42001?
ISO 42001 provides a framework for managing AI responsibly, including its risks, limitations, and human oversight. For us, that means being clear about how we use our proprietary AI models within a complete molecular design service. We handle the design, chemistry, and simulation work for lipid nanoparticle and protein degradation projects, with people reviewing the results at each stage.
Start with the Right Inputs
A useful design starts with a clear brief and reliable data. We review the target structures, starting molecules, and available experimental results against your project's goals. Missing information and assumptions are recorded so they can be taken into account when we assess the candidates.
Review Every Design
AI gives us a starting point for molecular design. We then combine surface analysis, docking, and manual chemistry review to refine the candidates. We look at how each molecule fits, which interactions support it, and whether there is a practical route to synthesis. A promising AI score is one piece of evidence in that review.
Check Predictions with Simulations
We use molecular dynamics (MD) to examine how a design behaves over time, with FEP and MM-GBSA calculations to assess binding energies where appropriate. These checks help us refine and compare candidates. We explain the assumptions and limits of the results: a favorable simulation still needs to be tested in the lab.
Keep Decisions Traceable
We keep track of the inputs, model and workflow versions, and simulation settings behind our recommendations. Your candidate package includes the supporting results and our reasoning, so your team can see why a molecule made the shortlist. We review the final designs with you, and your team decides which candidates move forward to synthesis and experimental testing.