Skip to content

Cache Documents

Cache documents are Fabaio's working memory - a context caching system that stores intermediate results, summaries, tool outputs, and configuration data that Fabaio needs to reason over during conversations. This system enables Fabaio to maintain context, reduce redundant queries, and provide transparent visibility into how decisions are made.

1. What are Cache Documents?

Cache documents are context storage units that Fabaio creates, updates, or references during conversation execution. They represent the agent's working memory and provide a way to:

  • Store Intermediate Results - Save data retrieved from tools, APIs, or queries for reuse
  • Maintain Context - Preserve conversation context across multiple interactions
  • Reduce Token Consumption - Cache frequently accessed data to avoid redundant LLM calls
  • Enable Transparency - Provide visibility into what data the agent used to make decisions
  • Support Debugging - Allow users to trace agent decisions back to source data

Cache documents are essential for understanding how Fabaio processes information and reaches conclusions during complex multi-step operations.

2. How Cache Documents Work

2.1 Creation

Cache documents are automatically created when:

  1. Tool Outputs - When tools return data that may be needed later
  2. Large Data Sets - When query results exceed token limits
  3. Frequently Accessed Data - When the same data is referenced multiple times
  4. Intermediate Results - When processing multi-step operations
  5. User Requests - When explicitly requested for caching

2.2 Storage

Cache documents are stored in the context cache system:

  • Persistent Storage - Documents persist across conversation sessions
  • Indexed - Documents are indexed for quick retrieval

2.3 Retrieval

Fabaio retrieves Cache documents when:

  • Context Needed - When previous data is relevant to current queries
  • Token Optimization - When cached data can replace expensive queries
  • User Request - When users explicitly request Cache document access
  • Cross-Reference - When linking related information across conversations

2.4 Lifecycle

Cache documents follow a lifecycle:

  1. Creation - Document is created with initial data
  2. Updates - Document may be updated as new information arrives
  3. Reference - Document is referenced during conversation execution
  4. Retention - Documents are retained based on retention policies
  5. Expiration - Documents expire based on configured TTL (Time To Live)

3. Viewing Cache Documents

3.1 In Conversations

Cache documents are visible in conversation details:

  1. Conversation View - Open any conversation to see Cache documents
  2. Cache Documents Tab - Navigate to the "Cache Documents" tab
  3. Document List - View all documents created, updated, or referenced
  4. Document Details - Expand documents to see full content

3.2 Document Information

Each Cache document displays:

  • Document ID - Unique identifier for the document
  • Creation Time - When the document was created
  • Last Updated - When the document was last modified
  • Content - The actual data stored in the document
  • Metadata - Tags, source, and other metadata
  • References - Where and how the document was used

4. Benefits of Cache Documents

4.1 Performance Optimization

  • Reduced Latency - Faster responses by avoiding redundant queries
  • Token Savings - Lower token consumption by reusing cached data
  • Cost Reduction - Reduced API calls and LLM usage costs

4.2 Transparency and Auditability

  • Full Visibility - See exactly what data influenced decisions
  • Traceability - Trace conclusions back to source data
  • Debugging - Understand why agents made specific decisions
  • Compliance - Maintain audit trails for regulatory requirements

4.3 Context Preservation

  • Multi-Turn Conversations - Maintain context across conversation turns
  • Cross-Conversation - Reference data from previous conversations
  • Working Memory - Preserve intermediate results for complex operations

4.4 Collaboration

  • Shared Context - Share Cache documents with team members
  • Knowledge Reuse - Reuse cached data across different conversations
  • Team Learning - Learn from Cache documents in shared conversations

5. Cache Documents in Shared Conversations

When viewing shared conversations, Cache documents provide:

  • Full Transparency - See all data used in the original conversation
  • Validation - Verify conclusions by reviewing source data
  • Learning - Understand how agents process information
  • Debugging - Identify issues by examining cached content

Even in read-only shared conversations, Cache documents are fully accessible for review and analysis.

Note

Cache documents are a powerful feature for understanding how Fabaio works and making AI operations transparent and auditable. They enable users to trace decisions back to source data, optimize performance, and maintain context across complex operations. All Cache documents are subject to your organization's data retention and privacy policies.