Omniscope is decoding and indexing the immune system using generative AI, building structured representations of the receptor repertoires that record what a person’s immune system has encountered and how it responded. That record is extraordinarily informative and largely unreadable: sequencing produces vast numbers of receptor sequences, but mapping which receptor recognises which target, and what a given repertoire implies clinically, remains one of the harder open problems in computational biology.

Making the repertoire searchable would change both diagnosis and therapeutic design, since immune signatures precede symptoms in many diseases and receptor discovery underpins entire classes of treatment. The approach treats immunology as a large-scale representation learning problem rather than a series of individual assays. Backed since 2023 with an 8 million dollar seed round, Omniscope is an early, technically ambitious position in the portfolio where the value depends on whether the index becomes a resource that others build on. It is a long-horizon position, but a working index of immune recognition would be valuable to nearly every immunology program.