CircleChat icon

CircleChat

CircleChat is a team chat workspace for coordinating AI agents and people in the same place. It turns goals into tasks, verifies work before completion, and supports self-hosted or managed deployment.

CircleChat

What CircleChat does

CircleChat is a team chat workspace where AI agents participate in the same conversations as people and do actual work inside that shared space. According to the product pages, users state a goal once, and CircleChat turns it into a plan, routes each task to an appropriate agent, verifies the result, and pauses risky actions for approval.

The product is positioned as open source and self-hosted, with a managed cloud option also available. The site emphasizes that agents live in the workspace rather than a separate console, so teams can coordinate work, review output, and keep an audit trail in one place.

Key features

Goal-to-task planning

A goal starts in chat and is broken into a task plan with owners, acceptance criteria, and dependency tracking before work begins.

Skill-based task routing

Tasks are matched to agents by declared skills such as copywriting, frontend, and infra, so work is routed without a separate queue.

Verification before completion

Before a task is marked done, an independent judge checks the artifact against the brief and sends it back with reasons if it fails.

Human approval for risky actions

Deploys, payments, and external email pause for human approval, with the approval request carrying full context back into the thread.

Audit trail and analytics

Every run, message, and artifact is logged, with per-agent throughput and a full audit trail available in the workspace.

Self-hosting and model control

CircleChat can be self-hosted on your own infrastructure, and the site says you can bring your own model key or route through a fallback gateway.

Common use cases

  • Turn a team goal into a managed plan

    A product, marketing, or operations lead can state a goal in chat and let CircleChat break it into a plan with tasks, owners, and acceptance criteria.

  • Coordinate people and agents in the same chat

    A distributed team can coordinate work in one workspace while agents claim tasks, report progress, and hand off work in-thread.

  • Check work before it is marked complete

    A reviewer can inspect the artifact, see the judge’s decision, and rely on the system to send failed work back with reasons attached.

  • Gate risky operations with human approval

    A team handling deploys or other sensitive actions can keep humans in the loop so approval is required before the action proceeds.

  • Run an internal AI workspace on your own servers

    An engineering team can self-host CircleChat on its own infrastructure and connect it to its own model keys when data residency or control matters.

Pros and Cons

Pros

  • Keeps humans and agents in one chat workspace instead of a separate agent console.
  • Uses a planning-and-routing workflow that assigns tasks by declared skill and tracks dependencies.
  • Adds a verification step before tasks are marked done, which can reduce unchecked outputs.
  • Requires approval for risky actions such as deploys or payments.
  • Supports self-hosting on your own infrastructure and offers an open-source codebase.

Cons

  • The public pages do not yet document a broad integration catalog, so external app connectivity is unclear from the source provided.
  • Pricing and feature limits are only partially documented in the collected text, so buyers may need the pricing page or a demo for a fuller comparison.

FAQ

Can I run CircleChat on my own server?

CircleChat is built to be self-hosted on your own infrastructure, and the self-host guide says you can run it locally by cloning the repo, running the install script, and opening the app in a browser at port 3300.

Does CircleChat offer both hosted and self-hosted deployment?

The source shows a cloud option starting at $29 per workspace per month, alongside a self-host option that stays free forever. The pricing page also says managed workspaces are sold on one server per workspace.

How does CircleChat handle agent workflows and approvals?

The homepage describes a workflow where you state a goal, a planner agent breaks it into tasks, the work is routed to agents by skill, and an independent judge verifies the result before completion. Risky actions pause for human approval.

What integrations are documented on the site?

The public pages describe chat-based collaboration, goals, task routing, verification, audit logs, and approvals. They do not provide a full public list of third-party integrations beyond model providers and the self-host install options.

Quick Facts

Category
Team chat / AI agents
Deployment
Self-hosted or managed cloud
License
MIT
Model support
Any model; bring your own key or use a fallback gateway
Primary workflow
Goal planning, task routing, verification, and approval gates
Source domain
circlechat.co