AI Documentation Search and Interactive Decision Trees

Prev Next

This guide describes how to navigate documentation using AI-powered features and interactive decision trees. It also outlines the process for providing feedback to ensure continuous improvement of documentation quality and accuracy.

Eddy AI Search

Document360 utilizes Eddy AI, a generative search tool that provides immediate, contextual answers to technical queries. This tool reduces the time spent sifting through multiple articles by retrieving and synthesizing information directly from the knowledge base.

The Eddy AI search interface showing a query and a generated response with citations.

Workspace Search Scoping

Eddy AI searches the active workspace by default to ensure search results remain relevant to the specific product version you are viewing. This scoping reduces "search noise" by preventing legacy content from appearing in searches for the latest PCA solution.

Note: Searching Across Versions

To find version-specific material, such as Skylight orchestrator 23.12.4 Release Notes, you must navigate to the Skylight Documentation workspace before initiating the search. If you perform the search from the Latest Documentation workspace, these legacy notes will not appear in the results.

Watch the example video below to see this in action. Since we are currently asking a question within the Latest Documentation workspace, Eddy AI is excluding results from the Skylight Documentation workspace.

This targeted search approach is essential as the documentation library expands to include upcoming on-premise releases and integrated PCA components.

Access AI Search

To generate an AI-powered response, follow these steps:

  1. Ensure you are in the correct workspace where you want to search or find results.

    Note: Use the top navigation bar to switch between workspaces. You will be redirected to its homepage.

  2. Click the Search bar on the homepage or at the top of any article.
    Tip: Press Ctrl + K to open the search dialog instantly.

  3. Enter your question or keywords in the search field.

  4. Press Enter. Eddy AI generates a response in the language of your query. Each response includes citations; click a reference number to open the source article in a new tab.
    Tip: Click the Copy icon to capture the generated text for internal use.

  5. (Optional) Choose the Like or Dislike icon to provide feedback. If you choose Dislike, enter a comment to help the Technical Publications team refine the AI model.

Search Capabilities and Constraints

  • Contextual Memory: Eddy AI maintains context based on the last five queries in a session.

  • Conversational Flow: Use the Continue conversation field to ask follow-up questions without restarting the search.

  • PDF Indexing: The search engine indexes content within PDF files attached to or embedded in articles.

Interactive Decision Trees

Interactive decision trees provide visual, step-by-step pathways to resolve technical issues or complete complex configurations. These trees appear in global search results and on the dedicated Search Page.

An interactive decision tree showing multiple logic paths for troubleshooting.

Decision Tree Controls

Use the following controls to interact with a decision tree:

  • Path Navigation: Click buttons to advance to the next step or choose a specific logic path.

  • Selection: Choose options from dropdown menus when prompted for specific environment data.

  • Step Jumping: Choose the step navigation dropdown on the left to view or select any step in the sequence directly.

  • Reset and Share: Click the More options (…) menu to Restart the tree or Share the link via email.

  • View Mode: Click the Expand icon to enter full-screen mode.

  • Exit: Click Back in the top-left corner to return to the documentation portal.

Improve Content Through Feedback

Cisco encourages stakeholders to provide feedback to ensure documentation meets organizational standards for clarity and technical accuracy. This feedback loop allows the Technical Publications team to prioritize content updates and address documentation gaps.

Feedback Guidelines

Individual comments are limited to 1,000 characters. Avoid including sensitive customer data or internal-only IP addresses in feedback fields.

Article-Level Feedback

At the end of every article, choose one of the following options:

  • Yes (Like): Confirms the article is helpful. You can add an optional comment.

  • No (Dislike): Signals a quality issue. You must choose a reason, such as:        

    • Need more information

    • Inaccurate or irrelevant content

    • Difficult to understand

    • Missing or broken link

To receive a notification when the team updates the article, select the Notify me about updates checkbox and enter your Cisco email address.

Search Result Feedback

If a search yields no results, use the feedback form to describe the information you were seeking. This data helps identify content gaps for the product roadmap.

Contact and Support

For further assistance regarding AI implementation in documentation or to report systemic issues, contact the Technical Publications team at techpubs_skylight@cisco.com.

© 2026 Cisco and/or its affiliates. All rights reserved.

For more information about trademarks, please visit:
Cisco trademarks 
For more information about legal terms, please visit:
Cisco legal terms
For legal information about Accedian Skylight products, please visit:  Accedian legal terms and trademarks