Promotion of molecules: New AI platform will save drug makers months of work

A Russian startup has introduced a platform that combines artificial intelligence and molecular modeling tools for the early stages of drug development, from the formation of design hypotheses to the selection of molecules for synthesis and biological testing. The project is currently undergoing closed testing. Representatives of pharmaceutical companies and academic teams involved in the design of medicinal molecules are involved in it. According to experts, the technology can significantly reduce the work time of scientists in the early stages. At the same time, such tools do not directly accelerate clinical trials. For more information about the possibilities and prospects of the technology, see the Izvestia article.

A Russian platform for creating medicines

The Russian startup Ligand Pro, founded by experts from Skoltech, presented a platform for the development of drug molecules using artificial intelligence and molecular modeling methods. This is the first Russian solution that combines the design of molecules, computational experiments and analysis of their results in a single digital environment.

With this tool, which has no analogues in Russia, researchers can formulate and test hypotheses of molecular design, generate and optimize structures, predict their interaction with a target protein, evaluate physico-chemical properties and structural risks, analyze the patent environment and synthetic availability of compounds. Based on the results of a series of calculations, the user can create a short list of the most promising molecules for subsequent synthesis and experimental testing.

"With the help of the platform, we make advanced AI tools available to pharmaceutical companies and research teams in order to accelerate the development of new drug molecules, increase the efficiency of computational experiments and move faster from hypotheses to promising candidate molecules," said Marina Pak, co—founder and CEO of the startup, graduate student and graduate student of Skoltech.

On the platform, molecules, protein targets, collections of compounds, design hypotheses, and calculation results are interconnected. This allows researchers not only to run individual models, but also to maintain the logic of decision-making over several cycles of designing molecules.

— Today, many teams use separate programs for different stages of development: one tool is for docking (computer modeling), another is for predicting properties, and the third is for generating or analyzing molecules. The results have to be manually summarized and interpreted, and the context of why a particular structure was created often remains outside the computer system. We did not want to create a catalog of online tools, but a single molecular design environment. A researcher can formulate a hypothesis, associate a series of molecules with it, verify it using calculations, and then select candidates based on the totality of the data obtained," said Sergey Nikolenko, head of the chemoinformatics department of the project.

The platform is designed primarily for specialists in medical and computational chemistry, as well as for research teams of pharmaceutical companies and academic organizations working on the search and optimization of bioactive compounds.

Mouse vision: neural network will significantly accelerate the creation of drugs for the brain

How automatic behavior analysis of laboratory animals will simplify drug testing

The demand for AI solutions in the pharmaceutical industry

The platform is currently at the stage of closed testing. Representatives of pharmaceutical companies and academic teams involved in the design of medicinal molecules are involved in it. The team is currently collecting feedback on operational scenarios and the interface and continues to add new computing modules. After testing is completed, access to the system is planned to be opened to a wider range of users.

The platform affects clinical trials only indirectly. Its main benefits appear at an early stage, when scientists determine which molecules are appropriate to synthesize and then test in the laboratory, Marina Chumakova, a leading market expert at NTI Helsnet, told Izvestia.

— Here, AI is able to reduce months of sorting through options and weed out weak connections to expensive experiments. The timing of preclinical and clinical trials is determined by biology and regulatory requirements, so the algorithm cannot shorten them. But the more accurate the selection at the entrance, the higher the chance that candidates with better properties will reach the tests and fewer projects will stop at late stages," the specialist explained.

Product validation has not yet been completed, but the AI approach itself seems promising, and the possibility of significantly reducing the duration of the earliest stages of the search for new drugs is certainly of interest, said Evgeny Orachevsky, Vice President for Government Relations at Pharmasynthesis Group. However, such tools do not directly accelerate clinical trials — their timing is determined by the design of the study, the time required to show and confirm the effect, and regulatory requirements for evidence.

— According to open sources, the molecules created with the help of AI successfully pass the first phase, where safety is assessed, and the second, where effectiveness is checked, does not yet show advantages over traditional methods. AI is able to take on a significant part of the rough work, but in responsible tasks, the final result still depends on the expertise of specialists who verify its conclusions and correct errors. Nevertheless, we continue to closely monitor the development of such platforms and are ready to consider proposals for testing them in a controlled manner.

The development was presented as part of the Moscow Startup Summit, which is taking place on September 29-30, 2026 on the territory of Sberbank.