Who we are, what we protect, and how to reach us to scope a campaign.
LillixBio is Celesium AI's AI drug discovery platform — the computational architect of preclinical discovery, from a named target to a synthesis-ready build list. We are one product line inside Celesium AI, and we share infrastructure and adjacent capability with the rest of the group — concretely, not just in spirit.
The shared platform layer LillixBio runs on: authentication, multi-tenancy, customer isolation, orchestration and metering. We didn't build a separate version of this — we build on the group's.
The group's restricted-party and sanctions-screening product — the kind of capability CLEAR's clearance work is adjacent to. Not wired into CLEAR today; the group operates it as a separate product.
The group's prior-art and patent-search product — the same discipline freedom-to-operate screening depends on. A sister capability, not a live call from DECIDE or CLEAR.
The group's defensible-document machinery — the same category of engineering that turns a screening result into an auditable build list or a clearance report you can hand to a regulator.
None of the above are integrations LillixBio calls at runtime today. They describe what the Celesium group operates and builds toward, not what ships inside DECIDE, CLEAR or OPERATE right now.
Two different kinds of protection, applied to two different layers of the stack — on purpose.
The biosafety gate, freedom-to-operate attribution and tamper-evident decision record are the subject of pending patent applications. Never described here as "patented" — pending means pending.
The internal methods of the selection engine are held as trade secret and are not described here or in diligence materials. What we publish is what it does, not how it does it.
Open protein language, structure and design models are licensed in, not held as proprietary. We hold no position that would compete with the tools you already use to generate candidates.
If you run generative campaigns and can see the waste in your numbers, we should talk. Engagements are scoped against a target you name and measured against a control you set.
We will not ask you for confidential sequence data in a first conversation, and we do not need it to show you how the platform works.