On Tuesday, October 13, Atlassian's Cloud list prices rise between 5 and 10 percent across Jira, Confluence, Bitbucket and the company's product collections. The vendor points to a year of investment in artificial intelligence. Strip away the messaging and the meaning is plain: enterprise search has stopped being a feature bundled inside collaboration software. It is now a budget line item.
Confluence, for years a wiki with a search bar, is being repositioned as an “AI-powered knowledge platform,” according to Atlassian partner Corptec, which published the detailed breakdown. Standard plans for Confluence and Jira rise 7 percent for teams under 5,000 seats and 10 percent above that threshold; Enterprise plans rise 8 percent; the Service Collection Enterprise tier rises 10 percent.
Enterprise search is getting a price tag
The headline increase is only half the story. On September 1, Atlassian announced that a usage-based billing model begins on December 3, 2026. Rovo AI credits, automation steps, Assets objects, customer-service AI agent resolutions and Bitbucket usage will be metered against pooled allowances. Partner explainers put automation-step overages at $0.50 per 1,000 steps (Dev.to and SPK and Associates).
Rovo, Atlassian's AI search and agent layer, draws on what partners describe as the Teamwork Graph, reportedly containing more than 150 billion objects and relationships. The pricing strategy reveals the company's bet: the index of everything the company knows is the asset, and access to it is what customers will pay for.
Atlassian is not alone in charging for search separately. Microsoft 365 Copilot lists at $30 per user per month for enterprise ($21 for Business annual plans). Google's Gemini Enterprise Business plan runs $21 per seat per month, with Standard and Plus tiers from $30. Notion charges $20 per member per month for Notion AI on Business plans, with its enterprise search still in beta. GoSearch offers a Pro plan at $20 per user per month. Glean, Coveo, Sinequa and Guru still price by quote, a sign that large deployments remain bespoke (GoSearch).
The market behind the enterprise search price hikes
The numbers explain why vendors think they can charge. The enterprise search market was valued at $7.47 billion in 2026 and is forecast to reach $11.66 billion by 2031, a 9.31 percent compound annual growth rate, according to Mordor Intelligence. The firm cites AI-powered search, cloud deployment and conversational interfaces as drivers.
A related segment is growing faster in absolute terms: the AI-ready enterprise knowledge graph market is forecast to reach $6.55 billion by 2036 from $1.05 billion in 2026, with metadata and entity graph platforms accounting for 36 percent of 2026 demand, according to Future Market Insights.
Adoption data suggests the spending is grounded in real usage. A Templafy survey of 2,000 knowledge workers at firms with 100 or more employees in Britain and the United States, conducted by Sapio Research in August 2026, found that 95 percent of knowledge workers edit AI-generated documents. Agent adoption among workers with access rose from 31 percent in December 2025 to 66 percent in August 2026. Median document creation time fell from about two hours to about eight minutes across more than 29,000 sessions.
At Barclays, the bank's internal knowledge assistant, introduced in 2025, has been used by more than 16,000 colleagues for over one million searches, the bank said in its October 1 announcement alongside Anthropic. About half of the bank's developers are expected to use Claude Code by the end of 2026, according to partner-reported adoption plans.
The knowledge graph underneath
The competitive pattern is converging on the same architecture: index everything, then sell answers. Glean, which bills itself as the open enterprise search layer, maintains more than 275 connectors across enterprise apps and on September 29 became a launch partner for OpenAI's B2B marketplace, letting enterprise customers spend OpenAI commitment dollars on its product.
The pitch across vendors is strikingly similar: employees waste enormous time finding information. Glean's partner Incentro puts that at up to a third of the working day, a partner-reported figure that should be treated as directional rather than audited.
“The operating layer” is how Shambhu Nath Jha, a senior analyst at Fact.MR, describes what comes next: “Context engineering will become the operating layer for enterprise AI. Companies will not get reliable agents by improving prompts alone. They will need curated knowledge and retrieval structures that make company information usable at the moment of answer.” (Fact.MR)
Atlassian's price schedule is the clearest commercial expression of that idea so far. The value in enterprise software is migrating from the application where you type to the index of everything the company knows. Vendors that own the index can charge for access by the seat and by the query. The open question for buyers is the uncomfortable one: once the index lives inside one vendor's metered platform, switching costs are no longer about moving documents. They are about losing the map of how the company thinks.