# Insilico previews virtual-cell research that models biological age > Source: > Published: 2026-08-16 03:32:27+00:00 [Insilico Medicine on X](https://x.com/InSilicoMeds/status/2088318461566976111?ref=runtimewire) In that post, Insilico Medicine says it is introducing biological age into its virtual-cell work. Insilico describes a Virtual Aging Cell webpage and a multi-agent generation platform, although the posting date could not be established from the supplied materials. The work closely tracks the research interests of [Alex Zhavoronkov (@biogerontology)](https://x.com/biogerontology?ref=runtimewire), founder and CEO of [Insilico Medicine](https://insilico.com/main?ref=runtimewire). Insilico presents the material as a preview of a research direction rather than a documented commercial release. The idea fits Zhavoronkov's work unusually closely. Before founding Insilico in 2014, he worked in graphics processing at ATI Technologies and then moved into bioinformatics and aging research. Insilico's [investor-relations biography](https://ir.insilico.com/en/corporate-governance/?ref=runtimewire) lists two bachelor's degrees from Queen's University, a biotechnology master's from Johns Hopkins University and a PhD in physics and mathematics from Moscow State University. Whether biological age can improve virtual-cell predictions or drug-development decisions remains a central test for the concept. ### The post framed age as a model variable Virtual-cell systems seek to predict what happens when a cell encounters a genetic edit, disease process or chemical compound. Adding biological age would make the task harder and could make the models more useful. An age-aware model could let researchers ask whether a target becomes more relevant as a cell ages, whether a molecule produces different effects across biological-age states, or whether an intervention shifts an aging signature without pushing the cell toward an unsafe state. Those uses remain prospective until evidence shows that the simulations predict experimental results across unseen cells and perturbations. Insilico has already built pieces that could sit around such a model. In 2026, Insilico [launched PandaClaw](https://insilico.com/news/spjz8fzmb1-insilico-medicine-launches-pandaclaw-emp?amp=true&ref=runtimewire), a natural-language agent inside its PandaOmics target-discovery system. Insilico says PandaClaw can assemble multi-step analyses using scientific skills and bioinformatics tools, drawing on multi-omics datasets and biological databases. Insilico's [LabClaw system](https://insilico.com/news/tmikccj2f1-advancing-drug-discovery-from-automation?amp=true&ref=runtimewire) carries the agent model into physical experiments. Insilico says LabClaw uses collaborating agents and specialist modules to coordinate target selection, compound screening, cell culture, sequencing, quality control and analysis inside its Life Star automated laboratory. Human approval remains at critical operating points. Connecting an aging-aware virtual cell to those systems would create a longer feedback loop. Agents could generate a hypothesis, simulate age-dependent effects, select an experiment and return the wet-lab result to the model. The strategic value lies in deciding which experiments deserve physical resources, where errors are expensive and biological combinations multiply quickly. ### Zhavoronkov is assembling data, agents and laboratory feedback Separately, Insilico announced a May 26, 2026 [collaboration with Human Longevity](https://insilico.com/news/ps2bndbh61-insilico-medicine-and-human-longevity-an?ref=runtimewire) to develop foundation models for longevity science. That project combines Insilico's model-training and evaluation systems with de-identified multi-omics, imaging and longitudinal health data from thousands of people, according to the partners. The collaboration reflects Zhavoronkov's broader push into computational aging. Insilico also has a drug-development program against which its modeling claims can eventually be judged. Insilico's published Phase IIa study of rentosertib, a TNIK inhibitor for idiopathic pulmonary fibrosis, offers one clinical test of its AI-driven discovery workflow. The [published Phase IIa study](https://www.nature.com/articles/s41591-025-03743-2.pdf?ref=runtimewire) provides a clinical benchmark for Insilico's discovery process. Later-stage testing must determine whether the reported result translates into a clinically meaningful benefit. That clinical program gives Zhavoronkov's computational work a practical destination. Insilico is trying to connect increasingly detailed representations of disease and aging to molecules that can survive laboratory testing and human trials. Insilico's agents are intended to coordinate the steps between those stages. ### Virtual cells are becoming a data competition The research direction described in Insilico's post places it in an active field. [Recursion](https://www.recursion.com/news/since-its-inception-recursion-has-been-building-the-foundation-for-the-first-virtual-cell?ref=runtimewire) combines large perturbational datasets, phenomics, transcriptomics, automated laboratories and high-performance computing in its effort to build a virtual cell. [Tahoe Therapeutics](https://www.tahoebio.ai/news/tahoe-therapeutics-raises-30m?ref=runtimewire) is building a one-billion-cell dataset intended to map one million drug-patient interactions. [Turbine](https://turbine.ai/news/turbine-25-million-series-b-virtual-biology-pharma/?ref=runtimewire) uses laboratory feedback to train virtual assays for oncology and immunology research. Those efforts establish a demanding standard for Insilico. An aging-focused virtual cell would need longitudinal or age-stratified biological data, reliable predictions on experiments excluded from training, uncertainty estimates and a route back to physical validation. Biological age also requires a precise definition: different clocks and molecular measurements can assign different ages to the same sample. Zhavoronkov has spent 12 years assembling systems around computational drug discovery. Insilico operates target-discovery software, generative chemistry tools, scientific agents, automated laboratories and an internal clinical pipeline. The concept described in Insilico's post could connect aging biology across that stack. Its value will depend on whether Insilico can show that age-aware simulation changes a drug decision and produces a result that holds up in the laboratory.