Atlassian Intelligence and Jira Service Management in 2026
Rovo, Virtual Agent, and the Real Cost
“If you already run Jira Service Management, Atlassian's AI layer is the path of least resistance. Rovo comes bundled at Premium ($51.42/agent/mo), Virtual Service Agent adds usage-based overage. For Jira-native shops, this beats the evaluation cost of bolting on Moveworks or Aisera. For non-Jira shops, there is no reason to migrate for the AI.”
Pricing
| JSM Plan | Price/Agent/Mo | AI Included | Virtual Agent |
|---|---|---|---|
| Standard | $20.00 | None | Not included |
| Premium | $51.42 | Rovo + Atlassian Intelligence | 1,000 conv/mo included; $0.30/conv overage |
| Enterprise | Custom | Rovo + Atlassian Intelligence + advanced | Included pool + custom overage terms |
The October 2025 price uplift context: Atlassian increased JSM Premium pricing in October 2025. For existing customers, the uplift applied at their next renewal. The $51.42 rate includes Rovo and Atlassian Intelligence, which were not included at the pre-October prices, so the effective AI cost increase per seat is partially offset by what would have been a separate AI add-on.
Virtual Service Agent overage above 1,000 conversations/month is the key variable cost. An organisation with 100 JSM Premium agents (100 x $51.42 = $5,142/mo) that generates 5,000 assisted conversations per month pays an additional 4,000 conversations of overage, which is $1,200/month at the entry rate. At 10,000 conversations it is $2,700/month at that rate. Model this against your estimated conversation volume before committing to Premium, and see the volume-discount note below before you take those overage totals as final.
Jira Service Management is free for up to 3 agents, and Standard, Premium and Enterprise each support up to 100,000 agents. The AI layer is priced separately from seats: Atlassian publishes that JSM includes 1,000 Virtual Service Agent assisted conversations per month at no charge, and that conversations above that threshold start at $0.30 each with volume discounts applied. Assets follows the same shape: 50,000 objects free, then from $0.05 per object per month with volume discounts. Both took effect on 16 October 2024.
The wording matters and most comparisons drop it. Atlassian says these unit prices start at $0.30 and $0.05 with volume discounts, not that they are flat. So a high-conversation deployment should not simply multiply volume by $0.30 and present that as the bill. Take the tiering table from Atlassian in writing during the negotiation. Modelling flat is the safe way to set a ceiling, but it is the wrong way to compare against a per-resolution competitor that quotes a firm unit price.
The 3-agent free tier also changes the shape of a small pilot. A team under 3 agents can run JSM itself at no cost and pay only for the AI conversation volume it uses above the included 1,000 per month, which is a materially cheaper way to test an AI service desk than a per-seat commitment on any platform that has no free tier. The catch is that Virtual Service Agent and the rest of Atlassian Intelligence sit on Premium and above, so the free plan buys you the ticketing, not the AI.
What the AI Actually Does in 2026
Rovo Search: AI-powered search across Jira, Confluence, and connected third-party tools (Slack, Google Drive, SharePoint). Returns direct answers from source documents rather than just links. Useful for agent knowledge lookup and end-user self-service for informational queries.
Rovo Chat: conversational interface to the Rovo AI model. Agents can ask questions about their project data, generate draft responses, and summarise issue threads. End-users can use it for self-service via the JSM portal.
Rovo Agents: custom task-specific automation agents that JSM admins can build without code. Agents can be triggered by events (new ticket created, SLA breach detected) and execute multi-step actions (look up user context, fetch relevant KB articles, draft initial response). This is the agentic capability in the Atlassian suite.
Virtual Service Agent: the end-user self-service AI endpoint accessible via JSM portal, Microsoft Teams, and Slack. Handles conversational IT support requests, routes to human agents when AI confidence is insufficient, and learns from resolution patterns over time.
Implementation
For an existing JSM Premium instance, enabling Atlassian Intelligence and Virtual Service Agent is weeks, not months. Rovo Search and Chat activate at the plan level; no separate configuration is required. Virtual Service Agent requires intent configuration, which is JSM admin work: defining the use cases the agent handles, connecting KB articles, and setting escalation thresholds.
If you are migrating to JSM from another ITSM platform while also enabling AI, add 3-6 months for the ITSM migration itself. The AI layer is fast to enable; the migration is not.
Published Deflection
Atlassian cites 20-40% deflection in customer cases, notably lower than the figures Aisera and Moveworks are credited with, and more conservatively stated. There is no independent benchmark to place that range against: we checked on 7 September 2026 and found no publicly readable analyst-published AI service desk deflection rate, so the honest comparison is vendor claim against vendor claim. Our deflection rates audit sets out what each circulating figure turned out to be when we opened the page it is attributed to.
Who Atlassian Intelligence is For
Good fit
- +Existing JSM Standard or Premium customers
- +Organisations already on Confluence for KB
- +Mid-market teams wanting AI without new vendor
- +Teams upgrading from Standard to Premium for Rovo
Poor fit
- -Non-Jira shops (no reason to migrate for AI alone)
- -Enterprises needing deep agentic multi-system automation
- -High-volume environments where $0.30/conv overage adds up
- -Orgs with fragmented KB outside Confluence/Google Drive