Table of Contents


Using AI Features

Once an administrator has configured Magentrix AI Tools, three AI features become available to users: Semantic Search, AI Ticket Deflection, and the Wizard Assistant. Which features you see depends on what your administrator has enabled and the security role assigned to you.


Semantic Search

If Semantic Search is enabled for your portal, the existing search experience is enhanced to understand the meaning of your question rather than matching exact keywords.

  • You can phrase queries as full questions: How do I reset my password? works as well as password reset.
  • Results may include Articles and Wiki pages that do not contain the exact words you typed but address what you are looking for.
  • You only see results from content you are authorized to access.

AI Ticket Deflection

If AI Ticket Deflection is enabled, the AI offers suggestions whenever you start to submit a support case.

  1. Begin creating a case as you normally would.
  2. As you describe the issue, the AI analyzes the description and presents relevant Articles and Wiki pages that may resolve the problem.
  3. If a suggestion solves your issue, follow it and close the case form. Otherwise, continue to submit the case.

Suggestions are drawn from the Case Deflection Articles and Wikis your administrator has selected, and respect your access permissions.


The Wizard Assistant

If your security role has been granted access to the Wizard Assistant, a chat icon is available within the portal. Click the icon to open the chat window.

  • Predefined questions - Up to 4 starter questions appear at the top of the chat window. Click any one to ask it directly.
  • Free-form questions - Type your own question and press Enter. The Wizard Assistant answers using your portal's configured knowledge sources (Articles, Wikis, and Document Library files).
  • Citations - Where appropriate, the assistant includes references to the source content it used. Click a reference to open the original Article, Wiki, or Document.
  • Conversation history - The chat retains context within a conversation, so you can ask follow-up questions naturally.

The Wizard Assistant only references content you are authorized to access.


Opening and Display Modes

The Wizard Assistant opens from a button in the corner of the portal. Once open, you can resize the conversation to suit your task:

  • Minimized - the corner button, out of the way until you need it.
  • Side panel - opens alongside the portal so you can keep working while you chat.
  • Full screen - expands the conversation to fill the window for longer sessions.

Answers are formatted for readability - headings, bullet lists, tables, and code blocks (with a copy button) all render in the conversation, and the chat scrolls to the latest message automatically.


Attaching Files

You can send files to the Wizard Assistant for it to read and reference in its answer.

  • Attach with the paperclip menu, or drag a file onto the conversation (a Drop to attach prompt appears).
  • Supported formats include PDFs, images (PNG, JPG, GIF, WebP), spreadsheets (XLSX, XLS, CSV), and documents (DOC, DOCX, TXT, MD, PPT, PPTX).
  • Each file can be up to 25 MB, and you can attach up to 10 files per message.
  • Each attachment appears as a removable item before you send, so you can remove one you added by mistake.
  • Files are checked against the size and type limits as you add them, with immediate feedback if a file can't be used. Password-protected or encrypted Office and PDF files are not accepted.

Voice Input

Instead of typing, you can dictate your question.

  • Click the Dictate (microphone) button. The first time, your browser asks for permission to use the microphone - choose Allow.
  • Your speech is transcribed into the message box in your browser.
  • An optional step can refine the transcribed draft for clarity before you send it. You can always edit the text yourself first.

Reading Images in Records

When you ask about a record that has images embedded in a rich-text field - for example a screenshot pasted into a case description, a diagram in an Article, or an image in a long-text field - the Wizard Assistant can read and interpret those images as part of its answer, not just the surrounding text. You don't need to attach the image separately; it is picked up from the record content the assistant is already authorized to see.


Rich Answers

  • Video answers - when a video answers your question, it begins playing at the relevant moment rather than the start.
  • Shareable links - related links are shown as labelled cards you can open or copy.

Wizard Agent Actions

If your administrator has enabled Wizard Agent record actions and your role is permitted, the Wizard can also help you create, update, or delete records on the entities you already have access to. The Wizard Agent always operates within the limits of your security role - it can never act on records you would not be able to act on yourself.


Tips for Better Results

  • Be specific. "How do I export a deal report filtered to closed-won deals from Q3?" produces better results than "deal report."
  • Provide context. Mention the entity, the role, the time period, or the specific action you are trying to perform.
  • Follow up in the same conversation. The Wizard remembers earlier turns; a follow-up like "and how do I share that with my team?" will use context from the previous question.
  • Click citations. The Wizard includes references to the underlying Articles, Wikis, or Documents - those are the authoritative sources if you need exact wording or to share with others.

<< Configuring the Wizard Agent | AI Tools Troubleshooting >>

Last updated on 7/5/2026

Attachments