Knowledge Base

General definition

A knowledge base serves as the factual foundation for retrieval-augmented generation systems. Rather than relying solely on what the model learned during training, the agent searches the knowledge base for relevant passages and includes them in the prompt. The model then answers based on that retrieved content, which reduces hallucination and keeps responses grounded in authoritative source material.

A knowledge base may contain product documentation, policy manuals, support articles, medical guidelines, legal texts, or any other structured or unstructured content the AI needs access to. Content is typically chunked, converted into embeddings, and stored in a vector database for efficient semantic search.

• Content freshness: knowledge bases require a pipeline to ingest updated documents so the AI answers from current information
• Chunking strategy: the size and overlap of document chunks affects retrieval quality and how well the model synthesises multi-chunk answers
• Access control: enterprise knowledge bases often need per-user or per-role visibility restrictions applied before retrieval
• Hybrid search combining vector similarity with keyword matching improves recall for technical or product-specific queries

In the Ethora ecosystem

Ethora’s AI SDK includes native support for knowledge-base ingestion and RAG-grounded bot responses. You upload your documentation or data, it is indexed and made searchable, and agents answer from that content. The knowledge base lives in your deployment, not on a shared cloud, which matters for documents containing protected health information or financial data.

Combined with self-hosted LLM agents, a private knowledge base gives you a fully on-premises AI Q&A system: your data, your model, your infrastructure. The agent queries your knowledge base, the model runs on your hardware, and no query or document leaves your environment.