search-strategy

Multi-source enterprise search uses query decomposition to break down natural language questions into parallel, source-specific sub-queries, executing them simultaneously across connected chat, knowledge base, and project tracker sources to retrieve comprehensive results.

Natural language query decomposition identifies query type, extracts keywords, entities, intent signals, constraints, and negations to translate natural language questions into source-specific syntax for accurate parallel retrieval across enterprise sources.

Yes, you can use natural language questions directly. The system translates queries into source-specific syntax automatically, executing parallel searches across chat, knowledge base, and project trackers while handling ambiguity without requiring manual query languages.

During result synthesis, the system aggregates outputs from parallel searches, removes duplicate entries, and applies ranking algorithms to synthesize a final answer while providing fallback strategies when connected sources are unavailable.

When sources are unavailable, the multi-source search provides fallback strategies to ensure continued operation. The system synthesizes a final answer using available results from the remaining connected chat, knowledge base, or project tracker sources.