Atlassian just added two capabilities to Confluence that most teams currently handle through workarounds: native visual asset creation and live connections to AI agents from Lovable, Replit, and Gamma.
The visual creation feature lets Confluence users generate diagrams and images without leaving the platform. Previously, that meant jumping to a separate tool, generating the asset, downloading it, and pasting it back. On documentation-heavy teams, that friction compounds across dozens of pages per week.
The agent integrations are the more consequential addition. Lovable builds web apps through conversation, Replit is a browser-based coding environment, and Gamma generates slide decks from written prompts. All three can now act as agents inside Confluence - you can trigger a task in any of them directly from a Confluence page, with that page's context shaping what the agent produces. A technical specification can, in principle, kick off a code prototype in Replit or a client-facing deck in Gamma without leaving Confluence to do it.
The Gamma integration is the most immediately practical for teams that present regularly. Gamma's AI already writes slide content from research and briefs, so tying it to Confluence means a project spec can generate a presentation-ready deck without moving the source material anywhere first.
What Atlassian Is Building Toward
The strategic play is making Confluence a coordination layer for AI work, not just a documentation store. If your team already writes specs and plans there, running AI tasks from that context removes the copy-paste cycle across multiple tools.
The open questions are practical: how much page content do these agents actually read, and can they write results back to the page automatically? Both matter a lot for whether this works in real workflows versus controlled demos, and Atlassian hasn't answered either yet.
Pricing details weren't confirmed. Based on Atlassian's rollout pattern, Premium and Enterprise tiers will see these features first. Verify availability before building workflows that depend on them.