I built a project manager nobody has to update

What if we didn’t need to keep interrupting people’s work just to make them describe it?

It’s the bane of every worker’s existence: “Did you update the project management system: Monday, Asana, Jira, Trello, or whatever else we’re using?”

One thing I know about most project management dashboards is that they’re almost always out of date.

This is a tax every legacy SaaS system puts on the team.

That’s the deal: You do the thing, then you go somewhere else and type that you did the thing.

It sounds small, but it really isn’t. It’s a tax you pay at the exact moments when you’re most productive. And the people who are best at the work are often the worst at paying it.

So the plan drifts. Then the plan needs a meeting to repair it. Then the meeting produces commitments that don’t get written down either.

It’s a doom loop, and everyone has been in it.

AI is changing that.

In reality, the status was never missing. It was scattered.

It was in the meeting transcript where somebody promised a deliverable. In the Slack thread where the scope changed. On the calendar where the follow-up got booked. In the sent folder where the thing actually shipped. In the completed work sitting in a shared drive.

The business was already producing a continuous, timestamped record of itself. We were just asking humans to hand-copy a summary of that record into a second system, badly, once a week.

So I built a project manager that reads the original instead of waiting for the copy.

It pulls from the evidence the business already creates: meeting transcripts, Slack conversations, calendar activity, sent and received emails, completed tasks, and deliverables.

From those sources, it maintains one current picture: what changed, what got completed, what’s stuck, who owns the next action, and what needs my decision.

Two rules make it work.

When the evidence is clear, it updates the status and next actions on its own. I started with a human approval queue, but that became another job nobody wanted to do. So I killed it and added a feature that lets you undo, reassign, or remove any automatic update with one click.

When the sources disagree, the agent asks for clarification in Slack. But as the system has improved, those conflicts have become rare.

So how does this work in practice? Here’s an example.

A salesperson promises a proposal during a client meeting. The calendar shows the follow-up date. The proposal goes out by email. The team discusses revisions in Slack.

Today, someone is supposed to update three tasks by hand. Usually, nobody does. Then the next status meeting spends ten minutes reconstructing a sequence of events that four systems already recorded perfectly.

Let the AI do that work instead.

The commitment gets logged from the transcript. The send gets recorded from the email. The revision becomes the next action, assigned to the person who owns it, with a reminder.

The owner gets pulled in under two conditions: the evidence is ambiguous or the deadline is slipping.

That’s the right split. Those are the only two things that were ever worth an owner’s attention anyway.

Most project meetings are sixty minutes of asking people what they’ve been doing.

That’s a waste of precious meeting time.

Now, we have the system summarize itself. Every Monday morning, a scheduled agent produces a summary of everything we did that week and drafts the agenda for our meeting.

It gives us the current operating picture: what moved, what closed, what’s stuck, who owes what, and what needs a decision.

The status update disappears. What replaces it is the conversation we were supposedly having all along: the one about the three things that are genuinely stuck and the two decisions only I can make.

The rule I’d write on the wall is simple:

Your team should do the work. The project manager should update the project plan.

You don’t need to build what I built to test whether the premise holds.

Pick one active project and give AI four things:

Then ask it to produce one page: current status, completed work, commitments made, next actions with owners, and unresolved decisions.

Compare that page with whatever your project tool says today.

The gap between those two documents is what your current system is costing you, and it’s usually wider than anyone wants to admit.