Vector database

Retrieval-augmented generation grounds answers by retrieving an organisation’s own documents at the moment of answering and passing them to the model as context. It keeps responses current without retraining, and it makes each answer traceable to the specific source the system actually consulted.

Knowledge base grounding constrains an AI system’s answers to content retrieved from an approved knowledge source. The system does not draw on the model’s own parametric memory. The common implementation is retrieval-augmented generation: the system searches a designated knowledge base for relevant documents, then generates a reply from that retrieved content. Regulated industries treat grounding as a control, not a quality feature, because it creates a traceable link between an answer and a specific source document — the record an audit requires. Grounding quality is bounded by the underlying knowledge base. Outdated or inaccurate source content produces outdated or inaccurate answers, regardless of the model’s capability. OnClarity’s AI Knowledge Agent grounds responses in an approved knowledge base as part of a compliance posture covering SOC 2, HIPAA, ISO 27001, GDPR, and PDPL.

Grounding is the practice of constraining an AI agent’s answers to approved knowledge sources and live system data rather than to the model’s own training. Grounding is the main defence against hallucination, and it is what makes an AI answer auditable back to a specific document or record.

A knowledge base is the maintained source of truth that AI agents and human agents both answer from. Its accuracy sets a ceiling on the quality of every automated answer, which makes knowledge maintenance an operational requirement rather than a documentation project owned by nobody in particular.