Knowledge Base

The Knowledge Base is where you upload and organize the documents that power your AI avatars' responses. Museum AI uses retrieval-augmented generation (RAG) to find relevant information from your documents and provide accurate, cited answers.

How It Works

1. Upload documents — PDFs, text files, and other formats
2. Assign to exhibits — Organize documents by exhibit for scoped retrieval
3. Automatic processing — Documents are parsed, chunked, and embedded automatically
4. AI retrieval — When visitors ask questions, the AI searches your knowledge base for relevant chunks and uses them to generate responses

Document Organization

Exhibits

Create Exhibits to group related documents. For example:

• "American Revolution" exhibit with battle descriptions, timeline documents, and biographies
• "Ancient Egypt" exhibit with artifact descriptions, historical context, and archaeological notes

Namespaces

Each document chunk is tagged with its exhibit and organization, ensuring avatars only access knowledge within their configured scope.

Document Processing Pipeline

When you upload a document:

1. Upload — File is securely stored
2. Processing — Text is extracted and split into semantic chunks
3. Embedding — Each chunk is converted to a vector embedding
4. Indexing — Embeddings are stored for fast similarity search
5. Ready — Document is available for AI retrieval

Chunk Analytics

Museum AI tracks which document chunks are most frequently retrieved, their similarity scores, and which conversations triggered them. Use this data to identify gaps in your knowledge base and improve content.

Tips

• Upload comprehensive, well-written documents for best AI responses
• Keep documents focused — one topic per document works better than large mixed files
• Use exhibits to scope avatar knowledge and prevent cross-contamination
• Monitor chunk analytics to find and fill knowledge gaps