qm

QM, built by YC Software, is a multiplayer system for running AI coding agents across an entire company, available both in Slack and as a web app. Most agent tools are built around a single person using a single assistant, and stretching that model to a whole team quickly gets messy. QM is designed from the start for that shared setting: every employee gets their own isolated workspace where they can work independently, and they can also bring the same agent into shared Slack channels, group messages, and projects to collaborate with teammates.

Every person and every shared room gets its own separate memory, files, list of connected accounts, permissions, scheduled jobs, and a durable sandbox where the agent can install and keep tools between sessions. QM is not tied to one underlying AI system. It can run on top of Pi, OpenCode, Codex, or Claude Code, and an organization can choose which of these are available and switch between them, since they all plug into the same shared core.

With QM, people can search internal notes, email, documents, and databases together in one place, build small internal apps and publish them to the right group of people, learn someone's writing style from past messages and draft email replies on a schedule, work inside an existing code repository to run tests and open pull requests, and post ongoing updates to a shared project channel. Administrators set organization wide configuration and choose a security posture: a strict mode that pauses for human approval before nearly every tool call, an automatic mode that screens external data and tool results before they reach the model, or a more permissive mode with no pausing, though certain destructive actions like recursive deletes are always blocked regardless of mode.

The core is written in TypeScript on Node, using Fastify for its API, Bolt for the Slack integration, and Vite with Lit for the web interface. Deploying it for your own organization means creating a separate deployment repository that depends on the published QM package and runs the setup command, which walks through infrastructure, sign-in, connector credentials, and verification, with each deployment running inside your own cloud account.