Identity resolution
Names, synonyms, properties, and structure.
System walkthrough
From molecule to evidence dossier.
One investigation runs the complete research workflow. The simplicity belongs in the interface; the scientific chain stays visible underneath.
Molecule
PubChem
Literature
Extraction
Classification
Synthesis
Dossier
01
Resolve the molecule
PubChem supplies the identity, chemical properties, synonyms, and structure. Capela does not ask a language model to invent chemical identifiers.
02
Search the literature
A bounded vocabulary drives Tavily searches restricted to trusted scientific publication sources across activity, targets, mechanisms, and computational studies.
03
Read the strongest studies
Duplicate records are removed, relevance is scored deterministically, and a bounded set of the strongest papers is analyzed.
04
Extract evidence
Structured analysis records only findings supported by each retrieved paper while preserving the source paper ID.
05
Validate classifications
A deterministic layer catches category errors: docking remains computational, expression changes do not become direct binding, and reviews remain literature reported.
06
Synthesize with restraint
Activities, targets, and mechanism edges are built only from validated evidence. Unsupported conclusions become explicit gaps.
07
Build the dossier
Capela assembles eight inspectable sections and verifies that every evidence and paper reference exists before saving the result.
AUTOMATIC
What Capela does automatically
The user supplies only a molecule. Capela handles the complete evidence investigation without a chain of separate prompts.
Literature retrieval
Bounded searches, deduplication, ranking, and source-depth tracking.
Evidence analysis
Structured extraction, deterministic validation, normalization, and aggregation.
Scientific assessment
Evidence strength, consistency, mechanism support, conflicts, and gaps.