Knowledge Management

Chao AI uses a knowledge management system to store, retrieve, and use your documents, notes, and conversation history to provide more relevant and accurate responses.

How Knowledge Works

When you add a context (document, note, URL, or file), Chao processes it into searchable knowledge:

1. Ingestion — Your content is analyzed and summarized
2. Chunking — Long documents are split into meaningful segments
3. Embedding — Each segment is converted into a semantic vector for similarity search
4. Retrieval — When you ask a question, Chao finds the most relevant segments

Context Sources

Source	How It's Processed
Note	Stored as-is for quick reference
Document	AI-summarized, then split into searchable chunks
File	Text extracted, summarized, and chunked
URL	Web page crawled, summarized, and chunked

Conversation Memory

When memory is enabled, Chao remembers key information across sessions:

• Your messages are embedded and stored for future reference
• Chao's responses (substantive ones) are also stored for continuity
• Memory is scoped to your account and never shared

Enabling Memory

Memory storage is controlled by the ai_memory_enabled preference. When enabled, Chao builds a growing knowledge base from your conversations that improves response quality over time.

How Retrieval Works

When you send a message, Chao:

1. Analyzes your query to determine if knowledge retrieval is needed
2. Searches your stored contexts and memories using semantic similarity
3. Selects the most relevant pieces of information
4. Uses that context to generate a more informed response

Chao skips retrieval for simple greetings, creative requests, and preference changes — only searching when your question would benefit from stored knowledge.

Best Practices

• Add relevant documents as contexts to give Chao domain knowledge
• Enable memory for persistent cross-session learning
• Be specific in your questions — Chao retrieves better with clear queries
• Review contexts periodically and remove outdated ones