AI Task Managers vs Task Boards With AI Members

Search for an AI task manager and you will find two different kinds of product sold under the same words. In the first, the AI is a planner built into the app: you give it tasks with lengths, deadlines and priorities, and it decides when you should do each one, usually by placing them on your calendar and reshuffling as the day changes. In the second, the AI is a member of a shared task board: an assistant you already use, such as Claude, ChatGPT or Cursor, connects to the board, reads the work, does some of it, and writes down what it did. The first manages your time. The second manages the work itself. If your problem is “I never know what to do next”, you want the first. If your problem is “an assistant is doing real work and I cannot see what it did”, you want the second.

Why the same phrase covers two products

Task management has always had two halves: deciding what matters and when, and keeping track of what is being done. Older tools put a person in charge of both. When AI arrived, product makers attached it to whichever half they already served. Calendar and personal-productivity apps attached it to the deciding. Boards and trackers attached it to the doing. Both call the result an AI task manager, which is fair, but it means comparing the two on features alone is comparing a diary with a workshop.

Kind one: the AI schedules and prioritises for you

Structurally, these apps work like this. You enter tasks with extra information a plain list does not ask for: how long each will take, when it is due, how important it is, sometimes what it depends on. The app connects to your calendar. Its AI then fits the tasks into the free gaps, in an order it has worked out from what you told it, and it re-plans when something moves: a meeting runs over, a new urgent task arrives, you skip a block.

Motion, for example, describes itself↗ (opens in a new tab) as taking your projects and tasks, prioritising them and time-blocking them on your calendar, re-optimising the schedule automatically through the day. Reclaim describes scheduling tasks and recurring habits into your calendar by priority, flexing them around your other commitments. Other apps take smaller steps in the same direction, suggesting an order for today rather than booking time.

What defines the category:

• The unit is time. A task matters because of when it will be done, so length and due date are central.
• The AI lives inside the app. It is the product’s own feature; you do not bring it.
• The AI advises and arranges. It decides when; you still do the work.
• The main view is a calendar or a day plan, and the main user is one person, even when teams are supported.

Kind two: the AI is a member of the board

Here the AI is not a feature of the board at all. The board is an ordinary list of work items in lanes, shared by the people who work on them. What changes is that an assistant can join it. Over MCP, the open standard↗ (opens in a new tab) for connecting assistants to tools (what MCP is), the assistant signs in, reads the tasks, adds new ones, moves them between lanes and comments on them, the same way a colleague would.

That brings in questions a scheduling app never has to answer, because in a scheduling app the AI never changes anything but your calendar. On a board, an assistant can mark work Completed, rewrite what a task says, or file twenty new ones. So the board has to know who the assistant is, limit what it can do, and record every change against its name. An AI agent task board sets out what that takes in detail.

What defines this category:

• The unit is the piece of work. A task matters because of what it is and where it stands, so the title, the note and the lane are central.
• The AI comes from outside. You connect the assistant you already use, and you can change assistants without moving the work.
• The AI does part of the job: triage, drafting, small fixes, summaries of what moved.
• The main view is a shared board, and the main question is accountability: who did what, and were they allowed to.

The differences that actually matter

1. What the AI changes. A scheduler changes when you work. A board member changes the work record itself. The first is easy to undo by looking at your calendar; the second needs a history that names who did it.
2. Where the AI comes from. A scheduler’s AI is the vendor’s. A board’s assistant is yours, so the board stays useful when you switch from one assistant to another.
3. Who else is involved. A scheduler is mostly about one person’s week. A board is about a list several people, and several assistants, look at and change.
4. What you have to tell it. A scheduler needs durations and deadlines to be any good. A board needs clear titles and a sentence on what done means.
5. What goes wrong. A bad scheduler gives you a day you ignore. A badly set-up board lets an assistant close work that is not finished. That is why roles and a human in the loop matter on one and barely come up on the other.

Who each one suits

An AI scheduler suits you if

• Your trouble is fitting many small commitments around meetings, and your calendar is where your day really lives.
• You work mostly alone, or your team’s shared work is tracked somewhere else already.
• You are happy to enter durations and deadlines for most tasks, because that is what the AI plans from.
• You want to be told what to do next, rather than to hand the doing to someone else.

A board with AI members suits you if

• You already use an assistant for real work — code, drafting, triage — and want its work to land in one list instead of a chat transcript.
• More than one person, or more than one assistant, touches the same work.
• You need to know afterwards who changed what, and to limit what each assistant may do.
• Your work is ordered by what matters next, not by the hour it is booked for.

Plenty of people need both, and they do not compete. A scheduler can plan your Tuesday while a board holds the product backlog your team and its assistants work through. The mistake is expecting one to do the other’s job: a scheduler will not tell you what an assistant changed last night, and a board will not book your afternoon.

What the board kind looks like on fenbs

fenbs is the second kind, and deliberately only that. A board has four fixed lanes — To Do, Next Up, In Progress and Completed — and every task is a feature, an enhancement or a bug with a priority from 1 to 10. It has no due dates, no reminders and no calendar, so it will not schedule anything. What it does is let an assistant work the board as a member:

fenbs_whoami        → who it is acting as, and its role on each board
fenbs_get_context   → the notes you wrote about how you work
fenbs_list_items    → what is in Next Up
fenbs_update_item   → moves BUG-031 to In Progress, writes its plan
fenbs_comment       → "Fixed in 7f3e21a; tests pass"

The assistant connects with a browser sign-in and holds your role, narrowed further by the scopes you tick when you approve it: read, write and comment. Its changes are recorded as “Claude via” the person it acts for, so its work can be told apart from yours, and you can revoke it in Settings without signing yourself out. If you are starting out, how to use AI agents in project management walks through a first pilot.