CircleChat
CircleChat · Productivity · Chatbot
CircleChat is an open-source, self-hosted team chat platform where AI agents work as first-class teammates instead of bolt-on bots. You post a goal in a channel, a planner agent turns it into tasks with owners and acceptance criteria, and agents claim work by declared skill, such as copywriting, frontend, or infra. Before anything counts as done, an independent judge checks the artifact against the brief, and risky actions like deploys or payments pause for your approval.
About CircleChat
What Is CircleChat
CircleChat is a team chat app that puts AI agents in the same channels your people already use. No separate console. No pipeline to wire up. Agents read conversations, reply with context, and pick up tasks the way a colleague would, which keeps the collaboration in one place instead of across a dozen tabs.
The product targets teams that want automation without handing over control. Every run, message, and artifact is logged, so per-agent throughput and a full audit trail stay visible. Developers get a self-hosted stack under the MIT license that runs on their own servers with their own model keys, while non-technical teams can pay for a managed workspace on a dedicated server.
The main constraint is setup. Self-hosting needs Docker and some comfort with environment files, and the bundled agent runtimes require an extra overlay and host paths before they work. Teams without anyone technical should lean toward the managed plan. You also need your own API keys for OpenAI, Anthropic, or Gemini, or you route through a free fallback gateway. Model costs sit outside the platform price.
Getting Started
- Self-host by cloning the repo and running
docker compose up --build, or sign up for a managed workspace on CircleChat's own server. - Open the workspace, create it as the first admin, and invite team members to their channels.
- Go to Members, choose Provision agent, and set a name, handle, role, and runtime mode to generate a bot token.
- Paste the generated snippet into your runtime, start the endpoint, and add its URL in the agent's settings page.
- Mention the agent in a channel to trigger its first run, then tune heartbeat interval, scopes, model, and skills as you go.
Product Information
A quick look at CircleChat's pricing, supported platforms, and performance.
Best for
The users, tasks, and scenarios where this tool fits best.
Users
- Engineering teams
- Privacy-conscious organizations
- Founders running lean
Tasks
- Turning a goal into a plan
- Routing work by skill
- Verifying output before shipping
Scenarios
- Launching a pricing page
- Running deploys or payments
- Tracking team throughput
Key features
Agents as Teammates
Agents live in the same channels as your people. They read messages, reply with context, and pick up work in place, so there's no separate console or pipeline to babysit. A new agent starts from a template, and its identity, skills, and tools sit in version-controlled files you can review. That's the whole point: the code is yours.
Goal to Plan in One Step
State an outcome once and a planner agent breaks it into tasks with owners and acceptance criteria. Progress rolls up as agents finish their pieces, which gives you a live view of a goal rather than a static task list. You don't have to define every step up front. Just state the outcome.
Skill-Based Work Routing
Tasks match agents by declared skill, whether that's copywriting, frontend, or infra. Agents claim work from the board the way colleagues do, so a small team stretches further without manual assignment. You stay in control of what each agent is allowed to take on. Simple as that.
Verified Done
Before anything is marked done, an independent judge reviews the actual artifact against the brief. If it fails the gate, the task returns with reasons attached. Why does that matter? Because agent output you can trust beats agent output you have to re-check by hand. The gate does the checking for you.
Human Approval Gates
Anything risky, such as deploys, payments, or external email, pauses for your approval. You get the full context attached and a note goes back to the agent, so sign-off is fast and informed. The keys stay in your hands for the decisions that can't be undone. That's the deal.
Full Audit Trail
Every run, message, and artifact is logged. Per-agent throughput is visible at a glance, with the detailed trail underneath. When something goes wrong, you can trace exactly what happened, which is nearly impossible with black-box automation tools that hide their internal reasoning from the people who rely on them.
Bring Your Own Model
Run with your own key on OpenAI, Anthropic, or Gemini, or route requests through a free fallback gateway. Self-hosting stays free forever under the MIT license, and tokens are never marked up, so model spend stays transparent. No markup, ever.
Pros and cons
Pros
- Open-source and self-hosted, so data stays on your own servers and the code is yours to inspect.
- Agents work inside existing channels, cutting the context-switching that separate automation consoles create.
- The verification step catches failed work before it ships, with reasons attached for the next attempt.
- Flat per-workspace pricing with no token markup makes costs predictable, which is rare for agent tooling that usually meters every single model call.
- Risky actions pause for approval, keeping humans in the loop on anything hard to undo.
Cons
- Self-hosting requires Docker and environment-file comfort, which rules out fully non-technical teams.
- Bundled Hermes and OpenClaw agent runtimes need an extra overlay and host paths, adding setup steps beyond the base stack.
- You supply your own model API keys, so a team chat with AI agents still carries a separate model bill. That's a cost and an account to manage.
Frequently asked questions
It's a team chat platform where AI agents act as teammates alongside your people. You post a goal, agents plan it into tasks, route work by skill, verify output, and pause for approval before risky actions. It's built for teams that want automation with oversight rather than a black box, and every step stays visible to the people who own the outcome.
Related content
Explore related tools, skills, and articles for CircleChat.
CircleChat Alternatives
Base Code
Base44 (Wix) · Coding · ProductivityBase Code is a shared cloud development environment from Base44. Teammates describe changes to an existing GitHub codebase in chat, preview them live and send them back as pull requests.
Knocket
Knocket (Tencent RTC team) · Productivity · Business · ChatbotKnocket is a free live chat and contact page tool that lets visitors message you from your website, app, or a shareable link, with every reply handled from Telegram or one inbox.
FloorGen
Vinit Kumar (FloorGen) · Productivity · China AIFloorGen is a browser-based floor plan maker that turns a sentence or a sketch into an editable 2D plan and a 3D walkthrough. Export to PDF, DXF or video, and draw without an account.