How a team actually uses Gnosis

People ask for an API as if an endpoint were the product. Sometimes it is. For Gnosis, the product is a decision: which molecules to synthesise, which to drop, and which experiment to run next. An API can come later. A confident button that hides a bad ranking would be a worse launch than no button at all.

What an engagement looks like

A typical starting point is deliberately small. A team brings an existing series, a target profile, or a liability they cannot shake. Gnosis ranks and generates inside the combined constraints of binding, selectivity, safety, downstream response, and ADMET. What comes back is a Candidate Intelligence Codex: prioritised structures, the scientific rationale in writing, and experimental recommendations a lab can actually schedule.

That is less magical than "integrate our API into your pipeline by Thursday," and more honest about where the time goes. The valuable hour is the one where a chemist argues with a flag, not the one where a JSON payload arrives.

What you can ask it to do

• Evaluate a series you already have, including the compounds you secretly hope are fine
• Prioritise analogues before you spend the synthesis budget
• Surface liabilities in exposure, selectivity, and safety early enough to change the design
• Generate candidates against an agreed profile, then read why they were proposed via Gnosis Lens

Why this is not self-serve yet

Absolute potency, RNA binding, and safety flags are not equally mature, and several of them refuse to answer unless the applicability domain says they should. Shipping that as an unattended API would mostly automate overconfidence. Early work with partners is how we learn which questions the system should answer out loud and which it should decline. If you have a program and a decision due, the contact page is the door. A public endpoint is a later chapter, written only once the refusals are as reliable as the scores.