Mission Control is where you turn one-off tasks into automated workflows. Instead of prompting an expert every morning, you set up a Robot — and it does the work on its own, on schedule, every time. > You're the commander. Assemble a team, set the goal, let them run. ## The difference between an AI Expert and a Robot (Task Agent) An AI Expert is human-driven — you start the conversation, it works, you get the result. A Robot (Task Agent) can run the same way when you assign it a task manually, but it can also run on its own — triggered by a schedule, an event, or a Webhook — and deliver results without you watching. Mission Control is where you build and manage those Robots.  ## How a Robot works Every Robot runs through a **Pipeline** — a sequence of stages, each handled by a specialist AI Task Agent: 1. **Inspiration** — analyzes the context, duties, and available resources to generate an actionable brief 2. **Goals** — converts the brief into strategic, high-level goals with clear deliverables 3. **Tasks** — breaks goals into concrete executable tasks and assigns them to agents or tools 4. **Validation** — checks each result against expected outputs and quality criteria 5. **Delivery** — formats and sends the final output to the configured destination When you trigger a Robot manually, **Robot Host** steps in first — it has a short conversation with you to clarify the task goals, then hands off to the pipeline. Each stage comes with a default AI Task Agent that works well out of the box. If you need precise control over a specific stage, you can swap in a custom agent for that stage.  ## Creating your first Robot 1. Open **Mission Control** from the sidebar 2. Click **+ Add Task Agent**  3. In the **Basic** tab: give it a name, choose a workspace, set the work mode (**Autonomous** to run on a schedule, **On Demand** to trigger manually), and set who it reports to  4. In the **Identity** tab: describe its role and responsibilities, then assign the AI Experts and tools it needs. Click **Save Changes** when done — then open the Robot, click **Assign Task**, describe the task, and Robot Host will confirm before the pipeline starts.  > 💡 Match the resources to the role. If the Robot is responsible for sending emails, give it access to Postman. If it needs to research the web, add the search tool. A clear role description with the right resources is what separates a Robot that works from one that doesn't. > 💡 Not sure what to write? Click **Generate** — it will draft the role description based on the name you provided. ## Scheduling and triggers Robots can run: - **On a fixed schedule** — hourly, daily, weekly, or any custom interval - **Continuously** — always running, like a background monitor - **Manually** — you kick it off when you need it, via the dashboard or a Webhook  ## Integrations The **Integrations** tab connects a Robot to the outside world. There are two sections: **Messaging platforms** — paste credentials to connect the Robot to a messaging app. Supported platforms: WeChat, Telegram, Discord, DingTalk, Lark. Once connected, users can chat with the Robot directly from that platform. **API Access** — each Robot exposes an OpenAI-compatible Chat Completions endpoint. You can call it from any OpenAI SDK or tool to trigger the Robot programmatically.  ## What's next - [What Is Mission Control](/docs/en-us/mission-control/what-is-mission-control) — a deeper look at how Robots, Pipelines, and Task Agents work together - [Creating a Robot](/docs/en-us/mission-control/creating-a-robot) — walk through setting up your first automation - [Scheduling & Triggers](/docs/en-us/mission-control/scheduling-and-triggers) — schedule options, continuous mode, and Webhook triggers - [Progress & Results](/docs/en-us/mission-control/progress-and-results) — monitor runs, read logs, and access deliverables
Mission Control: Orchestrate Automation
feedback
反馈