An AI stack that designs
gene-therapy vectors.
Gene therapy's bottleneck isn't the gene — it's getting it to the right cell. ExoAI is an autonomous fleet of agents that reads the literature, designs candidate vectors, runs the simulations, dispatches the wet lab, and files the IP. Outputs today: engineered exosomes. Outputs next: programmable LNPs for any receptor, any tissue.
Each agent owns one step of the design loop.
Lipid librarian
Scans 4.2M ionizable lipids from PubChem & internal libraries. Ranks by predicted endosomal escape, pKa, and toxicity.
Tropism simulator
Runs coarse-grained MD on top candidates against virtual tissue barriers (BBB, hepatic sinusoid, tumor stroma). Scores per-tissue uptake.
Ligand designer
Designs targeting ligands for any receptor — peptides, aptamers, antibody fragments — and predicts conjugation geometry on the LNP surface.
Payload engineer
Optimizes mRNA / siRNA / CRISPR / protein cargo for stability, codon usage, and lipid:cargo ratio. Generates cassette designs.
Wet-lab dispatcher
Queues the top 12 designs/week to our Emerald Cloud lab. Returns formulation, encapsulation %, and zeta potential.
Patent drafter
Watches every promising hit, drafts provisional claims, and files through our IP counsel. Maps the freedom-to-operate landscape live.
Test results feed back into Read. The loop tightens every cycle.
Every gene therapy program reinvents its vector from scratch. We're building the stack that designs them — fast, programmable, addressable to any cell. Engineered exosomes are the first vectors out of the loop and are already in the clinic. AI-designed LNPs come next. The agents are the product; the vectors are what they ship.