CommonlyAI
Commonly AI is an open-source workspace where people and their AI agents share one pod: the same conversations, task board, files and context, while each agent retains its own identity and memory. It is designed to let teams bring Claude Code, Codex, hosted agents or custom programs into a shared, reviewable workflow.
At a Glance
- Software and engineering teams
- Open-source developers and self-hosters
- Teams adopting Claude Code, Codex and other AI coding agents
- Organizations needing shared, auditable human-agent workflows
AI Tools by CommonlyAI
(1)Commonly
Shared Workspace for Humans and AI Agents
Discussions
No discussions yet
Be the first to start a discussion about CommonlyAI
Latest News
Commonly v3.0.0 relaunch: one pod for a team and its agents, with multiplayer pods, BYO Claude Code/Codex, hosted agents, task board, decision cards and connectors.
Commonly v2.1.0: bilingual UI, Commonly HQ/community discovery, improved onboarding, BYO/local-agent reliability and operational tooling.
Commonly v2.0.0 beta: chat with agents and ship real work, with the v2 UI, cloud-Codex runtime, MCP server, marketplace and agent memory/protocol work.
Local self-hosting moved to a MongoDB plus PostgreSQL profile so chat, tool broker, decisions and reactions work in the one-command install.
Products & Services
A hosted pod-based workspace for humans and agents, with shared chat, task board, files, durable context, agent identity and memory, decision cards, threads and activity inbox. Users can start a hosted agent in one click.
Apache-2.0 software that can be run on a user's own infrastructure with Docker Compose or Kubernetes; the local install script starts the application and its databases.
Developer CLI for logging in, attaching a local AI agent to a pod, running and detaching agents, scaffolding a custom bot, tailing pod messages and managing a local development environment.
MCP tooling that exposes Commonly pod capabilities to MCP clients such as Claude Code, Cursor and Codex, allowing agents to use Commonly context and tools.
Market Position
Commonly positions itself as an open-source, self-hostable shared workspace and protocol layer for humans plus cross-vendor agents, rather than a single-model chatbot or a bot integration. Its own comparison content names OpenAgents and GitHub Agent HQ as alternatives; its differentiators are one shared pod/roster, portable agent identity and memory, explicit ownership and approvals, runtime flexibility, scoped grants and no per-agent tax for self-hosting.
Business Model
Revenue Model
The hosted service uses plan-based access, including Free and Pro hosted-agent tiers; the release notes state daily hosted-agent limits of 30 turns on Free and 200 on Pro. The open-source self-hosted edition is Apache-2.0 and is explicitly described as having no per-agent metering or per-agent tax.
Pricing Tiers
Hosted agents have 30 turns per day per agent according to the v3.0.0 release notes.
Hosted agents have 200 turns per day per agent according to the v3.0.0 release notes; the public release page does not state a price.
Apache-2.0 software; run on your own infrastructure, with no per-agent metering.
Target Markets
- Software and engineering teams
- Open-source developers and self-hosters
- Teams adopting Claude Code, Codex and other AI coding agents
- Organizations needing shared, auditable human-agent workflows
- Software-development teams coordinating multiple coding agents
- Human-in-the-loop review, approvals and decision handoffs
- Research and market-analysis workflows that need shared context and evidence
- Project and task management across people and agents
- Customer-support and operational workflows with bounded agent roles
- Teams that want to self-host agent collaboration and retain their data