Back
TechCrunchLaunchTechCrunch2026-04-09

Atlassian Brings MCP Agents Into Confluence — Lovable, Replit, and Gamma Launch as Built-In Collaborators

Atlassian rolled out Remix (open beta) — a tool that converts Confluence pages into charts, infographics, and scorecards in place — while simultaneously launching three MCP-based partner agents: Lovable (spec → UI prototype), Replit (tech doc → starter app), and Gamma (meeting notes → presentation). The agents go live inside Confluence on April 13.

Original source

Atlassian has shipped two complementary products in a single announcement: Remix, an in-page AI tool that transforms Confluence content into visual formats, and three MCP-based partner agents that work directly inside Confluence pages without requiring users to copy and paste content between tools.

Remix is in open beta now. It reads a Confluence page — meeting notes, project spec, strategy doc — and generates alternative representations: bar charts, timelines, scorecards, one-pagers. The output is embedded back into the page as a Confluence macro, keeping the document as the source of truth while making the data visually accessible. Atlassian describes it as "the equivalent of having a designer in every doc."

The MCP agent integrations go live April 13. Three partners are launching on day one:

- **Lovable** reads a product or feature spec from Confluence and generates a working UI prototype, including basic interactivity, returned as a Lovable project link embedded in the page. - **Replit** reads technical documentation or architecture notes and scaffolds a starter application in the language specified, with a live Replit environment linked from the page. - **Gamma** reads meeting notes or strategy documents and generates a polished presentation, returned as a Gamma deck embedded in the page.

The MCP architecture means the agents read the full Confluence page including metadata, comments, and version history — not just the visible text. This gives them enough context to generate outputs that are aligned with the team's working decisions, not just the final doc.

Atlassian's play here is clear: position Confluence as the control plane for the knowledge-to-output pipeline, rather than just the place where decisions get written down. If the integrations work as advertised, they could meaningfully reduce the "now go build it" friction that follows planning documents.

Panel Takes

The Builder

The Builder

Developer Perspective

Replit reading an architecture doc and scaffolding a starter app, linked directly from the Confluence page, is genuinely useful. If the output is good enough to be a real starting point rather than a throwaway scaffold, this closes a loop that teams waste hours on every sprint.

The Skeptic

The Skeptic

Reality Check

Confluence is where good intentions go to die in a company wiki. Adding AI agents that generate more outputs from those documents doesn't fix the underlying problem: most Confluence content is outdated, incomplete, or organizational theater. Garbage-in applies at enterprise scale.

The Futurist

The Futurist

Big Picture

Atlassian positioning Confluence as the MCP-connected hub for cross-tool agent workflows is a significant strategic move. If it works, every project document becomes a launchpad for automated execution — the gap between 'we decided this' and 'we built this' shrinks dramatically.