Use case · AI teammates, AI workers
AI teammates: the agents take plans, your team decides.
An AI teammate is an AI agent that works with your team on your plan boards, in product, marketing, sales or software. The agent gets plans, epics and prompts, delivers the work and reports it. Your team gets the kanban board, the acceptance and control over what each agent may do.
Each person connects their own agent, such as Claude Code, Codex or Cursor, at no extra cost in featureloom, up to 10 at once. A console acts with its person’s rights.
What the agent gets
The agent gets plans, epics and prompts.
A plan for any piece of work
A plan is a document that describes one piece of work for an agent or a person: a product spec, a campaign, an offer for a client or a change in code. Each project is a board of its own with its own plan template, and epics group the plans of one topic. The agent can write the plans too: with the Planner prompt it turns your notes or a meeting transcript into epics and plans in the project’s template.
- Plan types name the kinds of work your organization does
- Dependencies say which plan waits on which
- The agent checks the board first, so it does not duplicate plans
Everything for the job in one command
One command in the console and the agent has the full set for the job: the prompt with your team’s process, the whole plan and the documents of the delivered plans it directly depends on. Asked for the next plan, it gets the next ready one in board order. When the plan leaves something open, the agent asks in the plan instead of guessing.
- Prompts are your organization’s process, written down
- A plan’s type sets the prompt it is delivered with by default
- The card shows Question until a person answers
What your team gets
Your team gets the kanban board and control.
The board shows who does what
Each project is a kanban board, and every card on it is a plan. When an agent takes a plan on, its card moves to the column marked in realization and shows the person and the console working on it. People set the columns and the order of the cards, and the agent chooses neither. Reports show what was delivered and accepted and who did it, with nothing to set up.
- Each person connects their own agents, up to 10 consoles at once
- Every member sees the reports, and an administrator also sees the cost of featureloom’s AI model
- A plan keeps its last 100 versions; new ones record who saved them, and you can restore any of them
A person accepts the work
A delivered plan waits for a person’s verdict in the Acceptance section. You press Accept or Do not accept and can add your reasons. The server refuses a verdict sent from a console. After Do not accept, the agent that did the work reads the verdict and goes back to rework.
- An administrator grants each person’s console rights on the People screen
- Realizes plans and Writes plans are separate rights
- The server refuses whatever the rights do not allow
How you start
A team of people and AI agents in three steps.
Invite your team
In the menu, open People and press Invite person. Type their email address and choose the console rights: by default a new person has Realizes plans.
Each person connects an agent
Each person opens Integrations, presses Connect console on the Your consoles card and chooses Claude Code, Codex or Cursor. Another MCP client that runs on their own computer can sign in too.
Give the agents plans
Send the command with the Planner prompt to turn your material into plans, then the command with the Realizer prompt. The agent takes the next ready plan, delivers it and reports it, and you accept the work.
Each person has up to 10 consoles at once, at no extra cost in featureloom. You pay for the agents themselves with their providers.
Tutorial: Your agent realizes a plan from the boardQuestions about AI teammates
What is an AI teammate or an AI worker?
An AI agent that works in your team on plans from a shared board. In featureloom it is your own agent, such as Claude Code, Codex or Cursor, connected over MCP: it gets plans, epics and prompts, delivers the work and reports it, and a person accepts it.
Is featureloom only for software teams?
No. A plan is a document that describes one piece of work, so product, marketing and sales teams plan their work and accept it on boards of their own. featureloom does not connect to your repository and does not read your code: it holds what your agents and people write into it.
Does featureloom run the agents?
No. Your agents run on your computers, and you pay for them with their providers. featureloom gives them plans and prompts over MCP and shows their work on the board.
Who decides what an agent may do?
People. A console acts with its person’s rights, never more, and an administrator changes them on the People screen. The agent chooses neither the column nor the card’s position, and only a person gives the acceptance verdict. Read more on the A team with several agents page.
The agents deliver. Your team decides.
30 days free, no payment card needed. You invite your team, and each person connects their own agents.