How to Add AI to Your Jira Service Management Customer Support (2026)

Lead

Let your small support team handle 3x the volume without hiring. Atlassian Intelligence in JSM triages requests, drafts replies, and suggests automations while your team stays focused on customers who genuinely need human judgment. Here's how to enable it in about 30 minutes.

Quick answer:

• Enable Atlassian Intelligence at the org level (3 minutes)
• Connect your Confluence knowledge base to JSM (5 minutes)
• Set up the Virtual Agent with intents and Slack channels (15 minutes)
• Add AI-powered drafts to your automation rules (10 minutes)

Verified against Atlassian Cloud JSM Premium and Atlassian Intelligence on 2026-05-05.

Prerequisites

How ScaleIT handles this: See Jira Rescue

• Jira Service Management Premium or Enterprise plan: Atlassian Intelligence and the Virtual Agent require Premium tier or higher
• Atlassian org admin role: needed to enable Intelligence at the org level
• Confluence space with knowledge base content: at least 10-20 articles covering your most common support topics. If you don't have one, create a space and draft articles from your existing support tickets first (takes 1-2 hours)
• Slack workspace connected to JSM (optional but recommended): the Virtual Agent can handle requests directly in Slack customer channels
• ~30 minutes to complete the full setup

Steps

Step 1: Enable Atlassian Intelligence at the org level

All Standard, Premium, and Enterprise JSM plans include Atlassian Intelligence. Enable it so your space admins can use AI features:

In Atlassian Administration → Rovo → Rovo access:

1. Select your organization
2. Check Enable Rovo (Rovo is Atlassian's AI/automation layer; this activates Intelligence features org-wide)
3. Click Save

Atlassian Intelligence is now active for all apps in your organization. No further org-level configuration needed.

Step 2: Connect your Confluence knowledge base to JSM

The Virtual Agent and Atlassian Intelligence both pull answers from your Confluence knowledge base. Link them now:

In your JSM project → Project settings → Confluence knowledge base:

1. Click Link a knowledge base
2. Select the Confluence space where your support articles live
3. Click Link

Your JSM project now reads that Confluence space. The Virtual Agent will search these articles when answering customer questions, and Atlassian Intelligence will reference them when drafting agent replies.

Pro tip: If your articles are scattered across multiple spaces, create a single "Support Portal" space and copy or link the key articles there. Virtual Agent works best with a focused knowledge base.

Step 3: Set up the Virtual Agent in your portal and Slack

The Virtual Agent is a customer-facing AI chatbot. It handles first-contact triaging and knowledge-base answers before escalating to your team.

In JSM Studio → Agents → Create an agent → Virtual service agent → Channels & self service:

3a. Enable in the customer portal:

1. Toggle Portal to On
2. Under Knowledge base, confirm it shows your linked Confluence space
3. Toggle AI answers to On (this enables the Virtual Agent to summarize and answer from your knowledge base)
4. Click Save

The Virtual Agent is now live in your customer portal. Customers see a chat widget and can ask questions.

3b. Set up Slack intents (optional but recommended):

Intents are conversation flows that guide customers through specific requests (e.g., "Report a bug," "Reset password").

Still in Virtual service agent setup:

1. Click Create an intent
2. Name it (e.g., "Password Reset")
3. Add example phrases customers might say (e.g., "I forgot my password," "Reset my account")
4. Set Fulfillment to "Provide answer from knowledge base" or "Escalate to agent"
5. Save

Create 3-5 intents for your most common request types. The Virtual Agent uses these to match customer questions and either answer them or escalate them intelligently.

3c. Connect to Slack (if you use it):

In Virtual service agent → Channels → Slack:

1. Click Connect Slack
2. Authorize the JSM app to your Slack workspace
3. Select Request channel (where customer requests appear) and Agent channel (where your team works)
4. Toggle Virtual agent active to On
5. Save

The Virtual Agent now monitors your Slack channels. When a customer posts a request, it will attempt to answer from your knowledge base, or escalate to an agent.

Step 4: Add AI-powered drafts to automation rules

Atlassian Intelligence can draft responses to customers based on similar past tickets and your knowledge base. Automate this.

In JSM project → Project settings → Automation:

1. Click Create rule
2. Set Trigger to "Issue created" (or use another trigger)
3. Set Conditions to match request types you want to auto-draft for (e.g., "Request type is 'Billing Question'")
4. Under Actions, click Add action → Atlassian Intelligence → Draft a reply
5. Configure:

• AI context: "Use knowledge base and similar issue history"
• Tone (optional): "Professional" or "Friendly"
• Add the draft to a comment on the ticket (so agents review before sending)

1. Add a second action: Notify → Assignee (so the agent sees the draft)
2. Save and activate

Now when a matching ticket is created, Atlassian Intelligence automatically drafts a response and notifies the agent. The agent reviews it, edits if needed, and sends it. No more typing from scratch.

Optional second rule: Create another automation that drafts replies only for high-confidence matches (set a confidence threshold in the rule conditions, if available). This keeps agent review time low.

Step 5: Monitor and refine

For the first week:

1. Daily (10 minutes): Review Virtual Agent conversations in JSM. Check:

• Did it answer correctly, or escalate appropriately?
• Did intents fire correctly?
• Are customers getting good first-contact resolutions?

1. Weekly: Pull a report of auto-drafted replies. Check:

• Did agents accept, edit, or reject the drafts?
• Are there patterns in what the AI gets wrong?

1. Refine: If the Virtual Agent is escalating too much, add more Confluence articles or adjust intents. If drafts are off-topic, update your knowledge base to be more specific.

Verify

Before scaling, confirm:

1. Virtual Agent is active in your portal: Visit your customer portal in an incognito window and confirm you see the chat widget.
2. Slack integration is live: Post a test request in your Slack request channel and confirm the Virtual Agent responds or escalates.
3. Atlassian Intelligence drafted at least one reply: Trigger a matching automation rule by creating a test ticket. Confirm a draft comment appeared within 1-2 minutes.
4. No hallucinations: Scan 5 auto-drafted replies and confirm each cites (or references) actual Confluence articles, not invented information.

Troubleshooting

The Virtual Agent isn't responding in Slack: → Confirm the Slack integration is "Active" in JSM Studio (Channels → Slack). → Check that your Request channel and Agent channel are correctly mapped. → Verify the JSM Slack app has the right permissions: chat:write, users:read.

Atlassian Intelligence isn't drafting replies: → Confirm Atlassian Intelligence is enabled at the org level (Rovo → Rovo access). → Check that your JSM project has a linked Confluence knowledge base (Project settings → Confluence knowledge base). → Verify the automation rule's Action includes "Draft a reply" and a notification so the agent is alerted.

The Virtual Agent is answering questions it shouldn't: → Your Confluence articles may be too generic. Rewrite them to be more specific to avoid false matches. → Delete or consolidate duplicate articles that confuse the matching algorithm. → If an intent is firing incorrectly, edit its example phrases to be more specific.

Customers are escalating after a Virtual Agent response: → The response may be technically correct but not addressing the customer's real question. Update your Confluence articles to cover the "why" behind common requests, not just the "how." → Add more intents to better categorize different types of requests. Vague intents lead to poor escalations.

Call to Action

Enabling Atlassian Intelligence and the Virtual Agent is straightforward if you have a solid knowledge base and time to tune intents. Many small teams do this and then let it go stale, watching deflection rates drop as tickets change. ScaleIt helps Atlassian-focused SMBs build AI support that stays excellent long-term. We've configured 20+ JSM deployments with AI and know exactly where setups break and how to fix them fast.

Get a free 30-minute consultation: We'll review your current JSM setup, audit your knowledge base readiness, and outline a roadmap to 50%+ first-contact resolution with AI. No homework beforehand, just a conversation about where you are and where you want to be.