An AI project manager
Half of project management is asking people what happened and writing it down. This job builds the plan, watches the dates, notices what has slipped and sends the update. When something is genuinely at risk it tells you rather than quietly moving the date.
Concretely
- Creates project plans from the shape of the work you describe.
- Tracks deadlines across the tools the work actually lives in.
- Assigns tasks and follows up on the ones nobody has touched.
- Monitors progress and identifies delayed tasks before the deadline, not after.
- Sends status updates so nobody has to run a meeting to find out where things are.
The status meeting exists because nobody wrote it down
A weekly meeting where everyone says what they did is a reporting mechanism with a very high price. It exists because the information is scattered across tickets, messages and heads, and collecting it is nobody’s job. When collecting it is somebody’s job, done continuously, the meeting can become shorter or stop being about status at all.
It raises the risk, a person makes the call
Rescoping, moving a client date or pulling someone off a job is always a person’s decision. This is the boundary that makes an automated project manager safe to run: it can see that a dependency has not moved in nine days and say so loudly, but it cannot decide that the deadline was optimistic. Those decisions have consequences it cannot see.
Delay is visible early or not at all
Projects rarely fail on the deadline; they fail three weeks earlier when a task quietly stops moving and nobody notices because everything else is progressing. Identifying delayed tasks is a daily comparison between the plan and reality, which is exactly the kind of work that gets skipped precisely when a project is busy enough to need it.
The jobs this comes from
Sanaf holds thirty five written job descriptions. These are the ones behind this page.