Give it work

Handing work to Player One is less like prompting a chatbot and more like asking a colleague. You describe the outcome you want, and Player One decides how big the answer needs to be: a reply, a task done here with you, a slice pushed to a helper, or a team of agents working on it in the background while you do something else.

You never have to choose the machinery. You do get to see it, approve it and stop it.

A quick answer

• You say: "What did we agree with the Halton account on payment terms?"
• Player One does: searches what it remembers about you and your files, and answers.
• You see it: the reply in the chat.

Nothing starts, nothing runs in the background. Most of what you ask for stays this small.

A task done here, with you

• You say: "Draft a reply to this enquiry, keep it under 200 words, and quote 2,400."
• Player One does: writes it in front of you and adjusts as you push back.
• You see it: the reply in the chat, and anything it makes appears in the preview pane on the right as it is made.

This is the shape to want when you care about the wording and expect to steer. You stay in the loop at every turn.

Side work, kept out of the way

• You say: "Write that page, and dig up what the three competitors charge while you are at it."
• Player One does: keeps writing the page with you, and starts a helper in its own conversation to do the pricing research. The helper reports back into your chat when it has an answer.
• You see it: a card when the helper is briefed and another when it replies, plus a row of faces beside the message box, one for each linked conversation. Click one to read the helper's own chat.

A long conversation is a slower and more expensive conversation, so the noisy part of a task is better done somewhere else and summarised back. Agents, delegation and team chat explains why that pays.

A hand-off to a specialist

• You say: "Get the researcher on this one."
• Player One does: starts the same kind of background conversation, but with the agent whose job it is.
• You see it: the same cards and faces. The full team is on Manage > Agents (/agents).

A specialist is worth reaching for because it keeps its own memory of its domain, so it gets better at that work the more of it it does. Your roster is yours to extend: see Agents.

A job, with a whole team

• You say: "Research what our four competitors charge, work out where we sit, and write me a report I can take to the board."
• Player One does: creates a job. A coordinator plans the work, splits it into tasks, picks the right agent for each and dispatches them. You get control of the conversation back straight away.
• You see it: Automations > Jobs (/jobs). Open the job for its Team, Tasks, Plan, Files, Messages and Activity tabs. The board updates live as tasks move.

The coordinator plans and delegates; it does not do the work itself. Files the team produces live in the job's own shared workspace, on the Files tab, and a job belongs to one project. Jobs are the right size when the work has genuine parts, needs more than one skill set, or will outlast the conversation. Jobs covers the parts, and Create your first job walks one through.

Approving before it runs

• You say: "Plan it with me before anything happens."
• Player One does: the coordinator writes a plan and holds the phases you gated until you decide.
• You see it: the plan on the job's Plan tab, and a Review Plan button on the planned task's card on the Tasks tab, with Approve and Reject. Rejecting asks you for a reason, which goes back to the coordinator so the next plan is better.

A job that runs into a decision only you can make parks itself and asks. The question arrives as a card in the chat you started the job from, and the job shows as Needs input on the board until you answer. It waits rather than guessing.

Watching, and stepping in

Three places tell you what is happening without your asking.

• Automations (/automations) is one feed of everything running by itself: jobs, flows and scheduled tasks, newest first, with Active, Completed and Failed tabs. It is a view rather than a control panel, so acting on a card takes you to the item behind it.
• Manage > Active Agents (/manage/active-agents/by-project) shows the conversations running this second, with the reply moving in the card as it is written. Stop agent interrupts a run without deleting the chat or the work it had already done.
• Messages (/activity), reached from the notification bell or the dashboard, is the cross-job feed of agents talking to each other, so you can read the team's own traffic when you want to.

Stop, Resume and Close job sit on the job page, and Stop asks you to confirm. Stopping a job pauses it rather than destroying it: resuming brings back its tasks, its team and its workspace exactly as they were. Stopping also returns before the team has actually stopped, so trust the board over the confirmation. A job that looks frozen is sometimes waiting for capacity from your AI provider and will pick itself up again, which is the one case where the right thing to do is nothing.

When the work should keep happening

Everything above is one-off. When you find yourself asking for the same thing every Monday, that is a different shape: a scheduled task, or a written process that runs the same way every time. See Make it repeat.