PerturbAI emerged from stealth and publicly released an in‑vivo Perturb‑seq CRISPR atlas mapping gene function across 8 million mouse brain cells, generated with partners 10x Genomics, NVIDIA and the Allen Institute. The startup also closed an oversubscribed pre‑seed round and said the dataset—designed to model causal genomics in intact tissue—will fuel AI training and therapeutic discovery. PerturbAI founders Xin Jin and Grace Zheng said the atlas reveals context‑dependent gene effects across cell types and regions and will support simulations of circuit‑level interventions. One‑sentence clarification: Perturb‑seq combines pooled CRISPR perturbations with single‑cell RNA profiling to link gene perturbation to cellular responses. Why it matters: a public, large‑scale causal genomics resource in intact brain tissue could accelerate target validation, enable better translation of preclinical findings, and provide foundational data for generative AI models in neuroscience drug discovery.