Source Attribution
Source attribution is the practice of returning, with every AI answer, a reference to the exact source passage used to construct that answer so the user can verify it.
Definition
In retrieval-augmented AI systems, source attribution is the mechanism that ties each generated answer back to the specific passages it was derived from. Implementations range from simple document-level citations to paragraph-level or even sentence-level highlights. Well-designed attribution lets a user (or auditor) click through to the source, verify the claim in context, and distinguish between information the system has grounding for and information it generated without source support.
Why it matters
Attribution is the difference between a convenient assistant and an auditable one. It makes the system's answers falsifiable, which is a prerequisite for high-stakes enterprise use. Attribution also supports compliance obligations: regulators increasingly expect AI-assisted decisions to trace back to verifiable inputs, and the EU AI Act's transparency requirements include making the basis of AI outputs reviewable.
How Volentis.ai handles it
Every Volentis.ai answer includes a citation with the document name and the exact paragraph used. Employees can click through to view the source passage in context. The full query + retrieval + answer trace is also written to an audit log for compliance review and GDPR subject access requests.