GenBio AI

Company building AIDO, a multi-scale "AI digital organism" foundation-model system that simulates cells and tissues for drug discovery.

1. Core Product / Service

GenBio AI builds AIDO (AI Digital Organism), a multi-scale foundation-model system whose stated ambition is to simulate biological systems from molecule up through cell and tissue, rather than modeling any single layer in isolation [local: virtual_cell_video_comparison.md]. Its virtual-cell component, AIDO Cell, was released in August 2026 as part of the broader AI Virtual Cell wave, positioned alongside systems such as Xaira's X-Cell and Arc Institute's Stack [local: virtual_cell_video_comparison.md].

Unlike single-cell transcriptome-only models, AIDO is reported to output across multiple modalities — numeric gene-expression matrices plus image modes — reflecting the "multi-scale digital organism" thesis rather than a point solution [local: virtual_cell_video_comparison.md].

2. Target Users & Pain Points

• Pharma / biotech discovery teams using virtual-cell simulation to predict perturbation response (which compounds change which cell states) before committing to wet-lab screening.
• Academic / translational researchers wanting a general-purpose cell simulator rather than building bespoke single-task models.

Pain solved: wet-lab perturbation screening is slow and costly; a credible virtual cell turns "run a screen" into "run a forward simulation," collapsing cost and cycle time. This is the same thesis as recursion-pharmaceuticals's industrialized phenomics, but from a model-first rather than data-first angle.

3. Competitive Landscape

Data from [local: virtual_cell_video_comparison.md]; treat funding figures as directional.

4. Unique Observations

• The "digital organism" framing is the differentiator. Most virtual-cell competitors predict one output (gene expression, or morphology). AIDO's multi-scale claim — one system spanning molecule → cell → tissue — is a bet that a single integrated foundation model beats a patchwork of task-specific ones. It is the same architectural wager ai-for-science tracks across all of AI4S, applied to the cell.
• Model-first vs data-first. Unlike recursion-pharmaceuticals, which built an industrialized wet-lab data flywheel first, GenBio AI leads with a foundation-model architecture — a cleaner test of whether the "just scale the model" thesis holds in biology, where aifs-biology documents single-cell foundation models repeatedly underperforming simple baselines.
• Timing. AIDO Cell's August 2026 release lands alongside the first zero-shot 2026 Virtual Cell Challenge, so its claims will be judged against a public, baseline-anchored benchmark rather than self-reported leaderboards — a rare accountability moment for the field [local: virtual_cell_video_comparison.md].

5. Financials / Funding

• Funding undisclosed as of the 2026-08-24 local research pass [local: virtual_cell_video_comparison.md]. No verified round size or valuation is recorded; treat any external number as unconfirmed.

6. People & Relationships

• Founder: Eric Xing (邢波) — machine-learning researcher and Carnegie Mellon professor, previously the founding president of MBZUAI [local: 2026-08-24-deepopencode.md]. The week's research also probed whether investor 胡祖六 (Fred Hu) participated; this was not confirmed in the local record.
• Peers: Xaira Therapeutics, Arc Institute, Cellular Intelligence, recursion-pharmaceuticals.
• Adjacent: david-baker (protein-design pioneer — same AI4S frontier, different layer).

Sources

• [local] virtual_cell_video_comparison.md — AI Virtual Cell modality/company landscape (compiled 2026-08-24)
• [local] 2026-08-24-summary.md — daily research summary
• [local] 2026-08-24-deepopencode.md — DeepOpenCode session log (GenBio AI founding-team + Eric Xing queries)