The Command Center

The Command Center is the screen you land on when you open your workspace, and it's where you put Jarvis to work. It's built around one idea: you should be able to direct an agent two ways (conversationally and as durable, repeatable work) without leaving the page.

So the Command Center is three surfaces in one:

• A Mission board, tasks the agent runs, tracked from brief to shipped.
• A chat with Jarvis, for quick, exploratory back-and-forth.
• A live HUD, streaming progress, activity, and workspace stats.

Missions: Work That Runs, Reviews, and Repeats

A Mission is a task an agent runs for you, not a passive to-do card. The mission's description is the briefing: the exact instructions handed to Jarvis when it runs. That's the key difference from ordinary project management, the card is the prompt.

Missions live on a board and move across four columns as work progresses:

To Do → In Progress → Review → Done

Starting a mission

Click New Task to open the mission console, then:

1. Name it and write the briefing, describe the outcome you want.
2. Attach files if the agent needs them; attachments ride into the prompt on every run.
3. Choose how to kick it off: Create & Execute (run now), Create & Schedule (run later or on a recurring cadence), or Save to To Do (run it when you're ready).

When a mission runs, its card moves to In Progress and streams live, then lands in Review with the agent's output attached. Multiple missions can run at the same time, each streaming independently, the Command Center is a work queue, not a single reply.

The review loop

This is where Agently turns one-shot AI into dependable work. From Review you can:

• Comment & Re-run, leave feedback and the agent runs again with your notes as context. Iterate until it's right.
• Ship, accept the result and move the mission to Done.

Because the briefing and every revision are saved on the mission, you get a durable, repeatable unit of work you can re-open and re-run later, not a chat you'll lose track of.

Chatting with Jarvis

Docked on the right is the Jarvis chat panel, for conversational work, quick questions, exploration, or one-off asks that don't need a mission. Slide it open beside your board, or expand it to take over the screen.

The chat isn't a black box. As Jarvis works, you see:

• Token-by-token streaming of its response.
• A tool timeline, each action (search the web, draft an email, update a board) appears as a live status row.
• Delegation cards, when Jarvis runs a Skill or hands a heavy sub-task to a specialist sub-agent, you see it happen.
• Citations, facts pulled from your Brain show their sources.

In the composer, type @ to pull a task, page, or Brain item into your prompt as context, and / to invoke a Skill by name. See Chatting with Jarvis for how to get great results.

Watching Work Happen

The Command Center reads like a mission-control dashboard, updating in real time:

• Live stats in the top bar, total tasks, how many are running, how many shipped.
• An activity ticker along the bottom, surfacing recent mission activity.
• A mission-control rail that groups what's running, in review, and recently shipped.
• Step-by-step run logs inside an active mission, so you can follow the agent's progress at each stage.

You're never left staring at a spinner, you watch the agent work.

Run Now, or Schedule for Later

Any mission can run immediately or be scheduled: a one-time run at a set time, or a recurring cadence like a Monday-morning digest or a daily inbox review. Scheduled runs execute in the background and post their results back on the mission, so work keeps moving when you're away. Scheduled runs also appear on your Calendar.

Missions vs. Chat: Which to Use

Most teams use both, chat to figure out what they want, then a mission to run it reliably.

Next Steps

• Meet Jarvis, How the agent plans and acts
• Skills, Teach Jarvis your repeatable playbooks
• Chatting with Jarvis, Write prompts that get results
• Spaces, The boards and tasks behind Missions