Break down large, complex topics into smaller, more focused documents. Each document should ideally address a specific question, task, or concept.

• Benefit: Makes it easier for agents to find the exact piece of information needed without parsing excessive irrelevant content. This improves the accuracy and speed of retrieval.
• Action: Review long documents and split them into logical sub-topics. Check the Tokens column in the Knowledge Base list view for an indication of document size.

Organize content within documents using clear, descriptive headings (H2, H3, etc.) and logical paragraph structures. Use lists (bulleted or numbered) where appropriate.

• Benefit: Improves readability for humans and significantly helps AI agents understand the hierarchy, context, and key points of the content, leading to better answer relevance.
• Action: Ensure consistent use of headings and formatting across all documents.

Use standardized terms and phrasing throughout your Knowledge Base, especially for key features, processes, or concepts.

• Benefit: Reduces ambiguity and helps the agent correctly identify relevant content even if user queries use slightly different wording.
• Action: Create a simple glossary or style guide if needed and ensure authors adhere to it.

Outdated or inaccurate information degrades performance and user trust. Set a schedule for reviewing and updating documents.

• Benefit: Ensures agents provide correct and current information.
• Action: Periodically audit documents for accuracy, relevance, and completeness. Remove or update obsolete documents.

Avoid having multiple documents covering the exact same topic with slightly different information. This can confuse agents and lead to inconsistent answers.

• Benefit: Provides a single source of truth, leading to more reliable agent responses.
• Action: Consolidate overlapping documents or clearly link related articles, designating one as the primary source.

AI agents don't typically 'read' entire documents like humans during a query. They use retrieval mechanisms to find the most relevant sections based on the conversation.

• Benefit: Understanding this helps prioritize clear structure and focused content over just accumulating large amounts of text.
• Action: Focus on making key information easily identifiable within sections. While there's no strict 'too long', extremely large and unstructured documents might make it harder for the agent to pinpoint the best answer efficiently. Settings like 'Top K' in the agent configuration often control how many relevant chunks are retrieved.