Enterprise Knowledge Search
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.
Reported impact
Document360 raised baseline AI accuracy to 94%, reduced user dislikes by 30%, and achieved 2-4 millisecond retrieval times while keeping access controls and content synchronization simple.
Document360 kept its knowledge-base AI stack on MongoDB by pairing MongoDB Vector Search on Atlas with Atlas Search and Voyage AI instead of replicating data into a separate vector store. The platform already served as a single source of truth for enterprise documentation and relied on detailed content permissions baked into MongoDB. When the Eddy AI assistant began handling more complex multi-intent prompts, small wording changes frequently pulled the wrong context and undermined customer trust. Rather than duplicate documents and permissions elsewhere, Document360 stored vectors in Atlas beside the source content and used hybrid keyword plus semantic search to retrieve the right passages natively. Voyage AI powers the embedding side of the stack, while Atlas keeps access controls and search in one governed platform. This gives Eddy low-latency retrieval, more reliable answers, and a simpler operational model for enterprise documentation use cases.
Documentation content and permissions remain in MongoDB Atlas, Voyage AI generates embeddings for that content, Atlas Search and MongoDB Vector Search on Atlas run hybrid retrieval against the same governed corpus, and Eddy uses the retrieved passages to answer complex user questions with low latency.
The unified architecture boosted baseline response accuracy to 94%, reduced user dislikes by 30%, and ensured 2-4-millisecond return times.
“Slight variations in how a user phrased a question also often led to entirely different contexts being retrieved.”
“This resulted in inconsistent answers that eroded customer trust.”
“Using a separate vector database for AI use cases meant we had to replicate all the data and keep it in sync.”
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.
The official page lists 'Headquarters: London'.
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 21, 2026 ElevenLabs Before shifting from AI research to enterprise products, ElevenLabs’ legacy database lacked native free-text search, draining engineering time. MongoDB Atlas MongoDB Sample Jul 30, 2026 Mentra Mentra needed a backend that would let a small team iterate rapidly on AI-native smart glasses while storing fast-changing app data, multimodal notes, and retrieval-ready embeddings for user context. The goal is to fit around 80 to 90% of faces with one frame. MongoDB Atlas · MongoDB Voyage AI 