Semantic Search and Document Indexing

Note

Available in the Enterprise edition only.

Semantic search allows finding relevant documents by the meaning of a query, not just by exact word matches.

Use Cases

• Header search (live search). When typing in the platform’s header search bar, in addition to standard results (documents, people, workspaces), a «Semantic Search» section appears with contextually relevant documents. For example, a query like «payment terms for client» returns contracts and commercial proposals containing payment information, even if they do not contain the exact phrase entered.

• Journal search. Semantic search is also available on the global search page — results are displayed in a separate SEMANTIC group alongside the DOCUMENTS, PEOPLE, and WORKSPACES groups.

• Access control. Users only see documents in search results that they have access to. If a user is not a member of a workspace, documents from that workspace are not displayed.

• Поиск по коду и документации (для разработчиков). Через AI‑ассистент можно искать по содержимому GitLab‑репозиториев — исходному коду, README, конфигурациям. Это работает через MCP‑протокол и не требует от разработчика знания конкретных файлов или путей.

Indexing happens automatically: when a document is created or modified, it appears in the search index within seconds; when deleted, it is automatically removed from the index.

Semantic search configuration is described in the RAG Configuration section.