Write together in one shared workspace, where your co-writers and AI agents work side by side, and every AI contribution stays visible to the whole team.
multi brings people and AI agents into the same document. Agent profiles and task definitions become shared resources, too.
Co-write in real time. Live cursors, presence, and comments mean everyone sees the same document as it changes. No more merging copies or losing track of who wrote what because of switches between the writing tool an AI chatbot. Delegate and discuss comments with AI agents, right where the conversation happens.
Bring AI into the workspace as a shared resource: Users can define their agents together. The profile editor allows defining an agent's expterise and skills. Users benefit from various agent perspectives. Agents suggest and make edits directly in the document, and their responses appear in the familiar comment feature for all to see.
Give agents work through shared tasks. Everyone can see what each agent was asked to do and what it produced, so AI use stays a coordinated, transparent part of the team's process. Agents can pick up tasks independently or on user demand. It is also possible to turn a task into an AI function.
multi began as a research tool. We built it to study how teams actually integrate AI agents into collaborative writing, then evaluated it in a one-week study with 14 teams, learning how people fold agents into their existing norms of authorship, control, and coordination. One of the key findings showed, that multiple agents can be useful to gain different perspectives on writing.
multi is a free, open-source project. Read the code, open an issue, or contribute. Everything is developed in public on GitHub.
