Skip to main content
Copilot Mode puts RAUM’s full four-step pipeline in service of your support agents. When you open a ticket, the AI analyzes it, searches your knowledge base, fetches relevant data, and drafts a complete reply — all streamed directly into the Zendesk composer. Your agent reads the draft, makes any edits, and sends. You stay in control of every response; RAUM handles the research and writing.
Copilot Mode usage is unlimited on all plans. There are no per-reply charges, no token caps, and no limits on the number of agents using it.

What appears in the Zendesk composer

When you generate a reply in Copilot Mode, the response streams into the Zendesk ticket reply editor in real time. You see the text appear word by word — the same way you’d watch someone type — so you can start reading before it finishes. Once generation completes, the reply is fully editable. You can change any part of it, add your own context, or replace it entirely before hitting send. The generation uses everything RAUM knows about the ticket: the message content, the full conversation thread, Zendesk tags, custom field values, any matching procedures from your knowledge base, and live data from connected tools when relevant.

Toolbar actions

The Copilot toolbar sits inside the Zendesk ticket editor. Each button runs a focused AI operation on the text in the composer.

Generate

Runs the full four-step pipeline against the ticket and streams a complete reply into the composer. Use this as your starting point for any ticket.

Rewrite

Rewrites the current text in the composer while preserving the meaning. Useful when the draft is structurally right but needs a different phrasing.

Expand

Expands a brief note or short bullet points into a full, polished reply. Start with a rough outline and let RAUM fill it out.

Friendly

Adjusts the tone of the current draft to be warmer and more conversational without changing the substance of the message.

Fix

Corrects issues in the current draft — unclear phrasing, missing context, or structural problems — without a full rewrite.

Quick Reply

Generates a short acknowledgment or holding response for tickets that need an immediate reply while you investigate further.

Grammar check and translation

Two additional tools run directly on the text in the composer without triggering the full pipeline. Grammar check reviews the current draft for grammar and spelling errors and corrects them in place. Run it before sending any reply you’ve written or heavily edited yourself. Translate detects the language your customer wrote in and rewrites your reply in that language. RAUM supports 20+ languages:
English, Spanish, French, German, Italian, Portuguese, Dutch, Polish, Turkish, Swedish, Danish, Finnish, Norwegian, Russian
RAUM auto-detects the customer’s language from the ticket content. You don’t need to select a target language manually — just click Translate and RAUM handles the rest.

Ticket sidebar

The Zendesk sidebar panel displays enrichment data RAUM generates for each ticket. This information updates in real time as the pipeline runs and is visible without opening the composer.
FieldWhat it shows
Sentiment badgeUrgent, Negative, Neutral, or Positive — based on the emotional tone of the customer’s message
Category pillThe auto-classified ticket category (Billing, Technical, Account, etc.)
SummaryA concise AI-generated summary of the ticket’s key issue and relevant context
Confidence badgeHigh, Medium, or Low — how confident RAUM is in its understanding and resolution
A low confidence badge is a useful signal to spend more time reviewing a generated reply or to check whether your procedures cover the topic adequately.

Response cache

When a ticket matches a type RAUM has resolved before, the response cache can serve an instant reply without running the full pipeline. Cache hits complete in near-zero time and consume no tokens. In Copilot Mode, cached replies appear in the composer the same way generated replies do — you review and send as normal. The cache is applied automatically. You can view cache hit rates per ticket category in the analytics dashboard.

When to use Copilot vs Agent Mode

Use Copilot Mode when...

  • Your team handles complex, high-value tickets that warrant human review
  • You’re onboarding to AI-assisted support and want agents to build confidence in the output
  • A ticket category involves sensitive decisions (refunds, escalations, account changes)
  • You want agents to learn from AI-drafted responses over time

Use Agent Mode when...

  • You have high-volume, predictable ticket categories (order status, password resets, FAQs)
  • You’ve validated RAUM’s output quality with Copilot Mode first
  • Your procedures and confidence thresholds are configured and stable
  • Ticket volume is too high for agents to review every reply
You can run Copilot Mode and Agent Mode at the same time. A common pattern is to enable Agent Mode for routine categories and use Copilot Mode as the fallback for everything else. See Agent Mode for setup instructions.