AI project management when the whole team is you

I spent eighteen months moving between project management tools. Kanban boards, sprints, a system with dependencies and Gantt charts, a beautiful setup with seven linked databases. Each one lasted about six weeks before I stopped updating it and went back to a notes app.

The problem wasn’t discipline. It was that I was using tools designed to solve a problem I don’t have. Project management software exists because teams need to know what other people are doing. I am the only person here. There is nothing to coordinate.

What team tools actually solve

Worth being precise, because it explains why they feel wrong alone.

Status columns exist so other people can see progress without asking. Assignees exist because work has to be distributed. Sprints exist to create shared commitment. Estimates exist so someone can plan capacity across several people. Dependencies exist because your work blocks someone else’s.

Alone, every one of those becomes a ritual you perform for an audience of yourself. Moving a card from “In Progress” to “Done” informs nobody. It just feels like progress, which is the trap.

What a solo operator actually needs

Three things, and genuinely not much more:

One list of next actions. Not projects, actions. Things you could sit down and do right now.

A view of what’s blocked and on whom. Waiting on a client, a supplier, an approval. This is the list that quietly costs you money when you don’t have it, because nothing chases itself.

A reliable capture point. Somewhere every commitment lands the moment it’s made, so it isn’t living in your head or in an email you’ll rediscover in March.

That’s it. Everything past those three is optional, and most of it is procrastination with a nice interface.

The task-versus-project distinction

The single most useful idea in any productivity system, and the one that fixed my lists.

A task is one action. A project is anything requiring more than one. Most people’s task lists are full of projects disguised as tasks, which is why looking at them produces paralysis rather than action.

“Sort out the newsletter” is not a task. You cannot do it. It’s a project containing about fourteen actions, and your brain knows this, which is why you keep skipping past it. “Write the subject line for issue 12” is a task, and you’ll do it in four minutes.

This is where AI genuinely earns its place. Paste a vague project and ask for the concrete first three actions, phrased as things you could start in the next ten minutes. It’s good at this, it takes seconds, and it dissolves the exact friction that causes procrastination.

The setup I’ve settled on

After all that tool-hopping, embarrassingly simple.

One database in Notion with tasks, each tagged to a project, each with a next-action phrasing and a date if it genuinely has one. Two views: today, and waiting-on. That’s the whole system.

I use Notion because everything else already lives there, meeting notes, content calendar, client records, and a task next to its context beats a task in a dedicated app I have to remember to open. My Notion AI review covers the AI features specifically, and Notion vs Obsidian for solopreneurs if you want the alternative.

The tool matters far less than two habits: everything gets captured immediately, and dates mean something. A date you routinely ignore trains you to ignore all dates, which is worse than having none.

The weekly review, which is the actual system

Thirty minutes, same time every week. Without this, any system decays into a list of things you’re not doing.

• Clear the capture point. Everything into a project or the bin.
• Check the waiting-on list. Anything older than a week gets a chase. This alone pays for the whole ritual.
• Pick next actions for each active project. One each, not five.
• Look at what didn’t happen. Not to feel bad, to notice patterns. Something that’s slipped four weeks is either not important or genuinely blocked, and both need a decision rather than another week on the list.
• Set the week’s three priorities. Three. Not ten.

AI helps here too: paste your completed items and ask what actually moved versus what was maintenance. The honest answer is often uncomfortable and useful.

Where AI doesn’t help

Being clear about this, because the marketing suggests otherwise.

It can’t tell you what matters. It has no idea which client is about to leave or which project you’ve lost interest in. Priority is judgement about your business, not a scheduling problem.

It can’t do the work. A perfectly decomposed task list is still a list.

And AI features bolted into project tools are mostly summarisation of things you already wrote. Useful across a team of thirty. Alone, you already know what’s in there.

If you’re wondering what to hand off more broadly, the AI stack that replaces your first virtual assistant covers where automation genuinely substitutes for a person, and how I automated 80% of my solo business admin with AI is the practical version.

My take

Use the simplest thing that holds a list and shows you what you’re waiting on. If that’s a text file, use a text file. The elaborate setup is almost always a way of feeling productive without producing anything, and I speak from eighteen months of experience.

Spend your systems energy on the two habits that matter: capture everything immediately, and do the weekly review. Those two carry a solo business fine. Everything else is optimisation of a coordination problem you don’t have.

Use AI at exactly two moments, breaking vague projects into first actions, and the weekly honest look at what moved. Both take under two minutes and both remove real friction. The rest of the time, close the tool and go and do the next action on the list.