Enterprise Knowledge Search
Kovai saw knowledge-base users shift from keyword search toward conversational questions, and needed to deliver low-latency, context-aware answers from documentation without introducing brittle data synchronization between its system-of-record database and a separate vector store.
Reported impact
Kovai delivered a seamless keyword-plus-AI search experience for Document360 users, returns vector results in under four milliseconds, reduced architecture complexity by avoiding a second vector store, and improved trust by grounding answers in relevant knowledge-base articles.
Kovai extended its Document360 knowledge-base platform with Eddy, a RAG assistant built on MongoDB Atlas and MongoDB Vector Search. Because the company was already using MongoDB as its system of record, the team chose to keep both articles and embeddings in the same Atlas deployment instead of syncing data to a separate vector database. OpenAI-generated embeddings are stored alongside knowledge-base content, and Vector Search retrieves the most relevant chunks by similarity so the assistant can answer questions with context, low latency, and linked source articles. The same application also preserves traditional keyword search, letting users switch between semantic and lexical retrieval in one experience. This unified architecture reduced operational complexity for engineering and data-science teams, improved answer quality, and gave Kovai a fast, cost-effective path to shipping AI-powered search directly inside a customer-facing SaaS product.
Documentation content is stored in Atlas, embeddings are generated and saved beside the source articles, Vector Search retrieves the best-matching chunks for a user question, and Eddy uses that context to answer while linking back to relevant knowledge-base articles.
Kovai adopted MongoDB Vector Search to unify content and embeddings, boosting performance, reducing sync issues, and simplifying architecture.
MongoDB Vector Search on Atlas allows Kovai to store both knowledge base articles and their embeddings together in MongoDB collections
Only answers from the knowledge base relevant to the context of the question are provided, and those answers are cross-referenced with relevant articles to ensure trust.
the team has seen average time taken to return 3, 5 and 10 chunks between 2 milliseconds and 4 milliseconds
Review how this reference was scored and trace each normalized claim back to its documented source.
Scoring methodology
Customer Reference Value measures usefulness as a customer reference. Evidence Confidence measures the quality and traceability of available evidence, not independent verification.
Traceability
Field coverage shows which facts each source supports. Dates are shown only when documented by the source.
Related
MongoDB Sample Sep 9, 2026 Squary AI Squary AI needed to eliminate slow, outage-prone legal search infrastructure that combined MongoDB with OpenSearch and hurt document review speed, cost, and trust. At Quatrini Law Group, Squary AI analyzed more than 460,000 pages and saved over 400 attorney and paralegal hours. Amazon Bedrock · Claude · MongoDB Atlas MongoDB Sample Archived · Aug 29, 2026 Document360 Document360 needed its Eddy AI assistant to answer multi-part documentation questions accurately without breaking access controls or adding a separate vector database to keep in sync. The unified architecture boosted baseline response accuracy to 94%, reduced user dislikes by 30%, and ensured… Atlas Search · MongoDB Atlas · MongoDB Vector Search on Atlas MongoDB Sample Archived · Aug 21, 2026 ElevenLabs Before shifting from AI research to enterprise products, ElevenLabs’ legacy database lacked native free-text search, draining engineering time. MongoDB Atlas 