
Provider of Customized AI Solutions for Drug Development
Biocompass (Beijing) Technology Co., Ltd.(Biocompass), the developer behind the AI Virtual Cell(AIVC) platform, has secured a new financing round totaling tens of millions of yuan, according to VCBeat. The round is led by China Reform Venture Capital Fund under national state-owned capital platform China Reform, with Dalton Venture and Yun Qi Partners joining as new investors. Existing backers FreesFund and Baidu Ventures also injected additional capital. Funds raised will primarily go toward developing a next-generation virtual cell algorithm model, collecting proprietary data and speeding up industrial implementation.
Unlike mainstream AI drug discovery tools that focus on molecular structure optimization, the AIVC platform covers the full lifecycle of drug development. Its capabilities span disease mechanism analysis, target identification, virtual screening of candidate drugs, preclinical efficacy and toxicity prediction, and patient stratification for clinical trials. Its core innovation lies in elevating AI modeling from the molecular level to the cellular level, enabling direct simulation of how living systems respond dynamically to external interventions.
Trained on massive single-cell multi-omics datasets, AIVC learns patterns of cellular state transition, gene regulatory networks and cross-scale causal relationships to build a foundational model that depicts cell behaviors and developmental fates. Researchers only need to input a cell’s initial status and specific perturbation conditions, and the model can predict subsequent changes involving cell proliferation, differentiation, senescence, apoptosis and disease progression. The platform allows researchers to conduct large-scale virtual experiments digitally and prioritize research hypotheses with the highest success rates for further validation via wet-lab tests. This shift from experiment-driven to model-driven R&D is expected to substantially cut costs, shorten development timelines and improve the overall success rate of drug research.
In September 2023, the founding team launched GeneCompass, a knowledge-enhanced multi-species large-scale cell model, which was featured as a cover story in Cell Research. Serving as a general foundational virtual cell model, it harnesses over 100 million single-cell data points and embedded professional knowledge. In 2025, the team rolled out CellGraphCompass, an upgraded model that systematically integrates graph algorithms and prior knowledge into virtual cell modeling. Leveraging its perturbation model built on the foundational framework, the team claimed first place in the preliminary round and second place in the overall ranking of a global virtual cell competition.
On the data front, Biocompass will build on a globally sourced AI-ready public dataset accumulated by its team over four years. The company will also develop in-house dataset technologies to gather high-quality time-series perturbation datasets tailored for AIVC training, and advance large-scale partnerships for clinical data collaboration.
In terms of commercial application, Biocompass has partnered with domestic and overseas players to deploy its models, co-develop cell therapy pipelines, identify early-stage drug targets and conduct precise stratification of clinical samples. The firm is building a complete industrial chain integrating models, data and real-world scenarios, and advancing coordinated development of advanced cell therapies and the combined AIDD and AIVC technology stack.