Audit Log
A tamper-evident record of queries and responses, used for traceability, governance and review.
Chunking
Splitting documents into smaller passages so they can be embedded and retrieved precisely.
Citation
A reference attached to a generated answer pointing back to the source passage it came from.
Compliance Gate
A checkpoint that documents must pass before they are indexed, enforcing governance rules.
Confidence Score
An indication of how well-supported an answer is by the retrieved source material.
Connector
An integration that lets Vijnara read content from a system such as SharePoint, Drive or Slack.
Data Residency
The requirement that data be stored and processed within a specific geographic region.
Embedding
A numeric vector representation of text that captures meaning for semantic search.
Grounding
Constraining a model's answer to retrieved, permitted context rather than its general training data.
Hallucination
When a model produces confident but unsupported or fabricated information.
Hybrid Retrieval
Combining semantic (vector) and keyword search to improve precision and recall.
LLM
Large Language Model — the model that composes natural-language answers from retrieved context.
Maker-Checker
A workflow where one person submits content and another reviews and approves it before it goes live.
Multi-Tenancy
Architecture that keeps each organization's data isolated within a shared platform.
Permission-Aware Retrieval
Retrieval that filters results by each user's role and department access before content is read.
RAG
Retrieval-Augmented Generation — retrieving relevant content and using it to ground a generated answer.
Reranking
Reordering retrieved passages by relevance before they are passed to the model.
Role-Based Access Control
Granting permissions based on a user's role, so access follows responsibility.
Semantic Cache
A cache of prior answers keyed by meaning, used to speed up repeated or similar queries.
Semantic Search
Search that matches on meaning rather than exact keywords, powered by embeddings.
Sub-Processor
A third party engaged to process data on the platform's behalf under contractual safeguards.
Vector Store
A database that indexes embeddings for fast similarity search across documents.