Here's the thing — you've spent an hour carefully writing up a project plan in your team wiki, only to copy the whole thing into PowerPoint anyway. Atlassian just said it's done with that. They've dropped a set of features that let documents inside Confluence transform directly into charts, prototypes, and even working apps.

TL;DR
Confluence AI visual tool Remix launches in open beta
Three third-party agents — Lovable, Replit, and Gamma — rolling out from 4/13
Built on MCP (Model Context Protocol) — any partner can connect an agent

What Is It?

Atlassian announced two things on April 8, 2026.

First, Remix with Rovo. Drag to select text on a Confluence page and AI instantly converts it into a visual — a chart, infographic, or scorecard. The key is that it doesn't touch your original document — the visual sits as a "layer" on top of the source page, automatically linked back to it.

According to Atlassian's own data, Confluence pages with visual elements get read nearly twice as often. Turns out, how well you present information matters just as much as how well you write it.

Second, third-party partner agents. Atlassian partnered with Lovable, Replit, and Gamma to bring AI agents that work directly inside Confluence.

  • Lovable agent: Product spec → working UI prototype
  • Replit agent: Technical docs → a starter app engineers can fork and use
  • Gamma agent: Meeting notes or status pages → a ready-to-present deck

These agents connect via MCP (Model Context Protocol). An admin enables the partner's MCP server in Atlassian Administration, and within minutes — no coding required — the agent shows up in the team's Rovo directory.

What Changes?

Until now, Confluence was "a well-organized document repository." Now it's turning into an execution platform where documents convert directly into deliverables.

BeforeAfter (Remix + Agents)
Spec → ChartCopy the doc → paste into Excel/Google Sheets → build the chart manuallyDrag text → Remix auto-generates a chart (linked to the original)
PRD → PrototypeHand off to a designer → days of work in FigmaLovable agent reads the PRD → working UI in minutes
Meeting notes → DeckOpen PowerPoint → copy content over → clean up the designGamma agent generates a presentation straight from the notes
Tech doc → AppFile a request with the dev team → sit in the backlogReplit agent auto-generates a starter app from the doc
The key difference: from doc to deliverable in one step, with zero context loss

Worth noting how this compares to something like Google NotebookLM. Atlassian's take: "NotebookLM works in a vacuum, but Remix operates within the pages, permissions, and structure your team already uses." The output isn't an isolated file — it stays permanently connected to the source document.

This isn't just a feature addition. After adding AI agents to Jira in February, Atlassian is now extending that to Confluence — putting their roadmap to agent-ify their entire product suite into full gear.

5M+
Rovo monthly active users
100B+
Teamwork Graph data points
More likely to be read when a page has visuals

Getting Started

  1. Check that Rovo is active
    Make sure you have Confluence Cloud and a Rovo license. Remix is rolling out in open beta to Confluence Cloud customers with Rovo enabled.
  2. Try Remix right now
    Look for the Remix button in the editor toolbar. Select a data-heavy table or a long document, then pick a suggested format — chart, infographic, or scorecard.
  3. Enable partner agents (admins only)
    Go to Atlassian Administration → Connected Apps and activate the MCP servers for Lovable, Replit, and Gamma. Rolling out from April 13.
  4. Call agents from Rovo Chat
    Open Rovo Chat on a Confluence page, select a partner agent, and give it a plain-language instruction — something like "Turn this spec into a Lovable prototype."
  5. Explore the MCP ecosystem
    Browse Atlassian's MCP skill gallery for additional integrations. You can also build your own agents.
Remix never overwrites your original. Feel free to experiment with different formats — if you don't like the result, just delete the layer.

Deep Dive Resources

Why "agentifying" work tools is happening now

Ever since Salesforce launched Agentforce as a standalone agent platform in 2024, the industry has been shifting fast — away from "build a new platform" and toward "embed agents inside the tools people already use." Slack turned Slackbot into an AI agent, and OpenAI chose to insert AI into existing workflows through consulting partnerships via Frontier Alliances.

Atlassian is following the same playbook. Agents landed in Jira in February, and now they've extended to Confluence. It all connects back to their decision to lay off 1,600 people (roughly 10% of the company) and redirect that investment into AI.

Why MCP is the real story here

The thing worth paying close attention to is MCP (Model Context Protocol). By choosing an open protocol over a proprietary API, Atlassian is saying any partner can build and connect an agent — no bilateral deal with Atlassian required. Lovable, Replit, and Gamma are just the opening act. As MCP adoption spreads, Confluence has a real shot at becoming a central agent hub.

Remix vs. Google NotebookLM

Both use AI to turn documents into other formats, but the crucial difference is workspace-native vs. standalone environment. Remix outputs live inside a Confluence page — team comments, mentions, and real-time edits all flow naturally alongside them. NotebookLM is more of a personal tool, and its outputs tend to get disconnected from the original source.