Build an Effective Knowledge Base

Premium KaiNexus Intelligence lets your organization create Custom Insights, Checkpoints, and Recaps tailored to your needs. When configuring these features, you can use a Knowledge Base to provide the AI with information specific to your organization.

A Knowledge Base can include your organization's definitions, standards, and procedures, helping the AI generate responses that reflect how your organization works. This article covers best practices for creating and maintaining effective Knowledge Base content.

Choose knowledge that supports the task

Start by considering what you want the AI feature to accomplish. Choose current, approved reference material that helps the AI complete that task accurately.

For example:

• Insights: Provide definitions or review guidance, such as the difference between cash savings and capacity released.
• Checkpoints: Provide the standards and exceptions relevant to the workflow review.
• Recaps: Provide terminology or organizational priorities that help explain activity and suggested next steps.

Keep the task and response format in the Prompt. Use the Knowledge Base for the reference material needed to carry out that task.

Each AI feature has its own Knowledge Base. Adding documents to one Insight, Checkpoint, or Recap doesn't automatically make them available to other AI features.

Organize the reference

When preparing documents for your Knowledge Base, use clear headings and direct language to make the information easy to understand. Depending on the content, you might organize your documents into sections such as:

• Scope: State which teams, Locations, processes, or situations the guidance applies to.
• Definitions: Explain important terms and use exact category or Field names.
• Rules: Describe the standards to follow and any supporting evidence required.
• Exceptions: Keep each exception beside its rule.
• Examples: Include a few examples that clarify difficult distinctions.

Include a document title, owner, version, or effective date when relevant.

Remove repetition and unnecessary details

Include only the information the AI needs to complete its task. Remove unrelated background, repetitive content, and unnecessary details that don't contribute to the AI's responses.

Keep your reference material concise without sacrificing accuracy. Preserve important information, such as conditions, measurements, dates, and exceptions, that could affect how the AI interprets or applies the guidance.

If documents contain conflicting guidance, confirm which information is correct before including it. Make sure the Knowledge Base doesn't contradict the feature's Prompt or Preamble.

Test and maintain the Knowledge Base

Adding reference material doesn't guarantee that the AI will interpret it correctly. Test your Knowledge Base using examples that include typical situations, exceptions, and missing information. Check that the responses follow your organization's guidance without inventing details or overlooking important rules.

If a response misses a rule or applies it incorrectly, check that the guidance is clear, relevant, and consistent with the other information available to the AI.

Assign an owner and keep the approved source documents. Review and update the Knowledge Base when your organization's policies, procedures, or definitions change, and test the updated content before using it.

Recommended Reading

• Write Effective Prompts and Preambles
• Create a Custom Insight
• Configure your organization’s Checkpoints
• Configure your organization’s AI Recaps