How to Create the Best Fact Structure: Naming, Descriptions & Content Input

Designing your custom stacks and facts in NOAN is the foundation for making your business AI-native. The way you name, describe, and structure your facts directly impacts how effectively the assistant can retrieve and use your knowledge. Here’s how to do it right.

1. Nomenclature: Naming Stacks & Facts

Stack Naming & Description

• Be Specific: Name stacks by business function or context (e.g., “Brand Strategy,” “Pricing Models,” “Sales Templates”).
• Describe the Scope: In the stack description, clarify what’s included.
  Example: “Contains all facts related to our brand’s market positioning, messaging, and visual identity.”

Fact Naming & Description

• Clear, Unambiguous Titles: Use concise, descriptive names (e.g., “Brand Positioning Statement,” “Outbound Email Template”).
• Purpose-Driven Descriptions: Write a one-sentence summary of what the fact is for and when it should be used.
  Example: “Defines our unique value proposition for B2B SaaS founders.”

2. Types of Facts & How to Structure Them

A. Informational Facts

(e.g., Brand Positioning, Pricing, Audience Segments)

• Title: State exactly what the information is.
• Description: Specify the context and intended use.
• Content:  
  • Use bullet points or short paragraphs.
  • Stick to one topic per fact.
  • Avoid mixing similar information (e.g., don’t combine “Pricing” and “Discount Policy”).
• Example:  
  • Title: “Brand Positioning Statement”  
  • Description: “Our core positioning for all external communications.”  
  • Content:  
    • “NOAN is the AI-native workspace for founders who want to scale with clarity and speed.”

B. Structural Facts

(e.g., Templates, Frameworks, Processes)

• Title: Name the template/process clearly (e.g., “Weekly Project Update Template”).
• Description: Explain when and how to use it.
• Content:  
  • Use numbered steps, placeholders, or markdown formatting.
  • Include instructions or examples where helpful.
  • Keep reusable elements modular.
• Example:  
  • Title: “Outbound Email Template”  
  • Description: “Use for first contact with B2B SaaS leads.”  
  • Content:  
  • Subject: Unlock AI-Native Growth for [Company Name]
Hi [Name],
[Personalized intro]
Here’s how NOAN can help you...
[Call to action]

3. Inputting & Maintaining Fact Content

• Structure First: Use headings, bullet points, and clear sections.
• Be Concise: Remove fluff—LLMs work best with focused, relevant content.
• No Overlap:  
  • Each fact should cover a unique topic.
  • If you find yourself repeating info, split it into separate facts and cross-reference if needed.
• Update Regularly:  
  • Review facts as your business evolves.
  • Archive or merge outdated facts to avoid confusion.

4. Crosslinking: Connecting Your Knowledge

Crosslinking facts—referencing one fact from another—can be highly effective for both human users and LLM-powered assistants.

• For the Assistant:
  When you reference another fact in a description (e.g., “See also: Discount Policy”), the assistant recognizes this relationship. If a user query touches on related topics, the assistant is more likely to pull in both facts, providing richer, more context-aware responses.
• For Users:
  Crosslinks act as navigational cues, helping you (and your team or advisors) quickly find related knowledge without searching from scratch.
• Best Practices:  
  • Use exact fact names when crosslinking.
  • Reference only the most relevant related facts to avoid clutter.
  • Update crosslinks if you rename or restructure your facts.

Summary:
Crosslinking is a smart way to guide both the assistant and human collaborators to the right knowledge, especially as your fact base grows. It’s not just for navigation—it actively improves the assistant’s ability to synthesize and connect information for more accurate, holistic answers.

5. Best Practices for LLM-Ready Facts

• Consistency: Use the same naming conventions and formatting throughout.
• Contextual Clarity: Make it obvious when and why a fact should be used.
• Atomicity: Each fact should be “atomic”—self-contained and about one thing only.
• Cross-Linking: Reference related facts in descriptions (e.g., “See also: Discount Policy”).

Bonus: Let the Assistant Help You Structure Facts

Whenever you’re creating or editing a fact—whether you’re in the fact block itself or developing it in the assistant workspace—you can ask the assistant to help you structure the information for optimal LLM readability. Just type your draft and prompt the assistant with something like:

• “Can you structure this fact for LLM clarity?”
• “How should I organize this template for best AI use?”
• “Suggest improvements for this fact’s naming and description.”

The assistant will analyze your content, recommend headings, bullet points, and crosslinks, and ensure your fact is ready for both human and AI consumption.

Note: Do not try to use the description field to try to influence or “hack” the assistant’s behavior. NOAN has safeguards in place to prevent descriptions from being used to manipulate model actions or outputs. Instead, focus on writing clear, natural language descriptions that simply explain the fact’s purpose and context. This ensures your facts remain transparent, maintainable, and effective for both humans and AI.‍

By following these principles, you’ll ensure your business knowledge is always accessible, actionable, and AI-ready—unlocking the full power of NOAN’s assistant for you and your team.

‍