Frequently asked questions

Despite advances in AI, no company has enough data to be able to build and train the transformative models needed to reliably get drugs from hit ID to phase 1 in less than 12 months. deepmirror builds the largest proprietary dataset for preclinical properties to develop a foundation model for drug design and unlock the promise of AI for Drug Discovery.

deepmirror is a well-trusted neutral party that helps its partners with drug development without developing its own assets. We only run laboratory experiments to improve AI models for our partners.

LLMs need specialised algorithms and data to effectively perform drug discovery tasks. deepmirror provides those algorithms and the data required to train them. Some of deepmirror’s functionality uses LLM integration, so as LLMs become more proficient we anticipate our product will improve as well.

With today’s models, we estimate that we can reduce the number of molecules we need to synthesise before dosing the first patient by 20%. But with more and better data, the models will improve further, and soon we’ll be able to develop drugs from hit to phase 1 in 6-12 months. One day AI will design the perfect drug for each patient.

Our foundation model will shrink preclinical drug discovery from the current standard 60 months to less than 12 months. For more detail on how AI adds value in drug discovery, read our blog on the topic.

Our platform, elio, already has 200 active users and our models are in use across more than 100 drug programs globally. In partnership with Medicines for Malaria Venture (MMV) and the Gates Foundation, we developed the Drug Design for Global Health (dd4gh) project, which has enabled the deployment of deepmirror’s technology in research laboratories across the world.

We have raised $5M from American and European funds, including SNR Capital, Twinpath, and Saras Capital. Our work has also been backed by grants from institutions including Innovate UK and the Gates Foundation.

elio is made for anyone in drug discovery and design, regardless of AI expertise. Our primary user base consists of computational and medicinal chemists.

elio has been built by chemists, for chemists; the platform is simple and accessible, even to chemists with no data-science or computational background. After uploading molecules and measurements, users receive generative design suggestions, and lab data feeds back into the models.

Speak to a member of the team for a full product demonstration.

elio is a subscription service, with flexible and milestone-based pricing depending on the needs of our customers. Get in touch with the team to find out more.

Yes, we have a number of impressive case studies detailing our work with customers. For example, after rebuilding its program using deepmirror’s combined binding model, Tes Pharma was able to run a billion-compound screen in just hours, and increased its hit rate from 8% to 60%. Read more about the results deepmirror can produce.

Sharing data is completely optional. You can choose to exclusively use deepmirror models that have been trained on publicly available data. However, opting to contribute data to the model’s development improves performance by roughly 20%.

No. All data is anonymised and aggregated, so no proprietary data is ever at risk. Read more about deepmirror’s stringent privacy and security policies here.