Atlassian says it is expanding its usage-based pricing model as it adds new AI capabilities. For IT admins, procurement leaders, FinOps owners, and SaaS operators, the operational question is not whether all AI becomes paid, but which interactions stay free, which become metered, and how limits are governed before spend starts to move.
What changed
Atlassian set out the change in a September 1, 2026 company blog post. Allowance limits and billing for the new meters go into effect on December 3, 2026.
Atlassian lists three meters in the announcement: Rovo credits, Automation steps, and AI agent resolutions. The company says Rovo credits cover deep AI interactions including Rovo Chat, Confluence AI Slides, Jira Coding Agent, and Teamwork Graph queries, while in-line summaries, rewrites, and Rovo Search remain free. AI agent resolutions are defined as outcome-based resolutions where an AI agent autonomously completes a request without human escalation. Those three are not the whole rollout. The same section notes that Assets availability is expanding and Bitbucket allowances are moving to org-level pooling, and Atlassian’s billing documentation lists both among the metered categories.
Why B2B teams should care
Atlassian positions the change as a way to monetize expanded AI use while still giving customers “full visibility and control over usage and spend.” That framing matters for enterprise buyers because it shifts AI cost management from a flat feature entitlement question to an active administration and governance task.
The near-term buffer, according to Atlassian, is that most customers on paid cloud plans already have built-in allowances. In practice, that means many teams may not feel the change immediately unless their Rovo Chat, Jira Coding Agent, or Automation usage pushes past those included limits. Atlassian says teams that need more can upgrade a plan, pre-pay for credits, or pay as they go, and that alerts arrive before limits are reached.
Allowances are not uniform across meters, which is the detail budget owners are most likely to miss. Atlassian’s Rovo documentation prices extra usage at $0.01 per credit, or $10 per 1,000 credits, above per-plan allowances that are pooled at the organization level. Customer Service Management is the exception: Atlassian’s support documentation states those plans do not include an allowance for AI agent resolutions and charges $1.00 per resolution from the first successful resolution once usage charges begin on December 3.
Atlassian also says its internal benchmarking found that agents with Teamwork Graph context delivered 44% better answer quality using 48% fewer tokens than agents without it. That is an Atlassian benchmark rather than independent validation.
Who is affected
The change applies to Atlassian cloud customers on eligible plans, especially those adopting Rovo, Automation, and AI agent features. Atlassian refers to “eligible plans” without listing every affected plan.
External reporting from The Australian Financial Review adds one useful packaging detail: customers currently receive a fixed number of AI tokens called Rovo credits with subscriptions, and the number of credits differs based on the type of product and monthly plan. Atlassian’s own documentation puts numbers on that packaging, listing per-user monthly credit allowances that rise by tier across Jira, Confluence, and its Service and Teamwork collections.
What teams should check now
Teams should review current usage patterns in the workflows Atlassian has identified as billable or potentially billable after December 3:
- Rovo Chat usage
- Jira Coding Agent usage
- Automation volume tied to metered steps
- AI agent resolutions, billed per successful resolution
Atlassian says organization and billing admins can adjust limits at any time and see real-time usage across teams in Atlassian Administration. Budget owners should confirm which allowances apply to each meter, define who is allowed to change limits or enable extra usage, and decide whether internal chargeback or approval controls are needed before meters go live. For the wider pattern, see our guidance on AI cost management for agentic AI and on how SaaS pricing models shift the cost drivers buyers must forecast.
What remains unclear
- Not yet confirmed: a published rate for Automation steps — Atlassian lists it as a meter, but its billing overview gives no per-unit price, unlike Rovo credits and AI agent resolutions.
- Not yet confirmed: the full list of “eligible plans” and whether all paid cloud tiers carry the same allowances.
- Not yet confirmed: how Assets objects and Bitbucket usage are metered and allowanced, beyond the expansion and org-level pooling noted in the announcement.
- Not yet confirmed: whether the no-allowance treatment of AI agent resolutions is specific to Customer Service Management or extends to other products, which the announcement’s “most paid cloud plans have built-in allowances” framing does not resolve.
What to watch next
The next concrete milestone is December 3, 2026, when allowance limits and billing for these meters take effect. Rovo credit and AI agent resolution rates are already documented; still missing are Automation step pricing, meter detail for Assets and Bitbucket, and confirmation of which plans count as “eligible.” Teams running Jira at scale should also fold Automation volume into that forecast, since step counts track workflow design rather than headcount — a driver absent from the seat-based math in our Linear vs Jira comparison.
