Atlassian's Year of AI: Did Rovo Earn Its Seat?

Atlassian's AI strategy in 2025 has been visibly more cautious than the rest of the productivity-software industry. Where Microsoft has built Copilot into the core productivity bundle and Notion has shipped AI-everywhere, Atlassian has positioned Rovo as a separately-priced add-on, structured around cross-product context (Jira, Confluence, Loom, the Atlassian platform broadly) and aimed at the "search across your company's knowledge base" use case.

The strategy is defensible. The execution, twelve months in, is mostly disappointing.

We tested Rovo at three Atlassian-deployed organizations over Q3 2025: a 280-engineer scale-up, a 110-employee professional services firm, and a 90-engineer SaaS company. Adoption was tracked against the per-seat investment.

Where Rovo wins

Cross-product enterprise search. The most useful feature in Rovo is the "ask a question, get an answer that synthesizes relevant Confluence pages, Jira tickets, and Loom recordings" workflow. The cross-product context is the architectural premise of Rovo and is also the part that delivers visible value. For organizations with significant Atlassian footprint, finding old decisions, surfacing related tickets, and summarizing project history are all meaningfully faster with Rovo than without.

The data integration is the second Rovo strength. The product respects Atlassian's permissions model (private Confluence pages stay private), the search index is genuinely up-to-date, and the response quality is reasonable.

The cross-product search is real. The rest is a defensive AI bundle priced as if it were a category-defining one.

Where the case is hard

Most Rovo features beyond enterprise search feel like checkbox responses to competitor pressure. The Jira ticket-summary feature is fine. The Confluence writing assistant is fine. The agent-builder tooling is technically present and functionally a step behind what ChatGPT Team or Claude for Business produce. None of these features, individually, justifies the per-seat price.

Adoption is the second issue. Across our three test sites, the percentage of seats using Rovo features more than weekly was, after three months, in the 18–28% range. The remaining 70%+ of seats were paying for the product without using it. This is a textbook adoption-failure pattern and is consistent across multiple Atlassian customers we have spoken with.

On the pricing

Rovo's per-seat add-on price stacks on top of Atlassian's existing per-seat pricing, which is itself meaningful at scale. The combined cost — Jira plus Confluence plus Rovo — is now in the $50+ per-seat-per-month territory for many enterprise customers. The math against alternatives is harder than the marketing implies. ChatGPT Team is roughly $25 per seat with materially better general-purpose AI capabilities. Claude for Business is in similar territory. The Atlassian-specific value of Rovo (the cross-product context) does not, in our reading, justify the premium for most organizations.

What to do

For Atlassian customers evaluating Rovo: pilot the cross-product search with a small group, ideally a knowledge-management-heavy team (engineering platform, ops, customer success). If the team uses it weekly and reports value, the product is worth the seat for that team. For most other teams, it is not.

For Atlassian to consider: bundling Rovo into the core product at the existing price tiers would be the right move. The current packaging produces visible adoption failure and is, in our reading, harming the AI-credibility story Atlassian wants to tell. The product is not bad. The pricing is.

The verdict

Rovo earns its 3.6 on the strength of the enterprise-search use case and the architectural premise. The pricing is the largest single criticism. We expect Atlassian to restructure the packaging in 2026. We hope it produces a different review.