Chemical data
PubChem PUG REST
About Capela
Scientific evidence, made easier to inspect.
Capela is an AI molecule investigation agent built for the Orion Agents Hackathon. It exists to answer one bounded question well: what does the available evidence actually support?
01
Why Capela exists
Molecule research is fragmented across chemical databases, paper indexes, full texts, and manual interpretation. Capela makes that workflow feel simple without hiding how conclusions were reached.
02
Evidence-first by design
Uncertainty is not treated as failure. Limited evidence, conflicting results, unsupported targets, and missing mechanisms are all valid investigation outcomes.
03
Built for a focused hackathon demo
The architecture is intentionally direct: a Next.js interface, a NestJS investigation runner, MongoDB persistence, and Privy authentication.
Scientific literature
Tavily over trusted publication sources
Structured analysis
Google Cloud Gemini with Groq fallback
Access
Privy authentication with free preview investigations
04
A clear product boundary
Capela does not perform docking, predict ADMET, generate molecules, recommend experiments, or decide whether a molecule is a good drug. It investigates the retrieved evidence and makes certainty inspectable.