From a Messy Brief to a First Plan

The brief you actually get

Textbook projects start with a clear charter. Real ones start with an email that says something like: "We need to get the new supplier portal live before the end of the year. Can you pull a plan together? Marketing want to be involved. Talk to Sam." Attached is a slide deck from a meeting you were not in, and a spreadsheet with four tabs, one of which is labelled "old DO NOT USE".

Your job in the first week is to turn that into something people can argue with. Not a finished plan. A first plan: what we think the scope is, the rough shape of the milestones, the assumptions we are making, and the questions nobody has answered yet.

This is one of the best uses of AI in the whole job. Structuring a mess is exactly what it is good at, and a first plan is supposed to be wrong in places, because its purpose is to surface the places where it is wrong.

Four sections, always

Whatever the brief, the first plan has the same four parts.

Scope. What is in, and what is out. The out list matters more, because that is where the arguments are hiding. If the brief says "supplier portal", is supplier onboarding in? Training? Migrating the old records?

Milestones. A handful of checkpoints, each with a test for whether you have reached it. Not dates yet, or at least not dates you believe. The shape comes first.

Assumptions. Everything the plan quietly relies on. Budget exists. Sam is available. The old system can be switched off. Writing these down turns hidden risks into visible ones.

Open questions. What you need someone to answer before the plan can firm up, and who that someone is.

The lazy prompt and the useful one

The most important line in the good prompt is the ban on durations. Left alone, a model will cheerfully fill a timeline with plausible numbers, and plausible numbers are the hardest kind to remove once people have seen them. Estimates come later, from the people doing the work. Lesson five has more on why.

The reusable prompt

Keep this somewhere you can find it. It works on a one-line email and on a forty-page document.

I am a project manager. Below is a brief for a new project, plus any notes
I have. Turn it into a first plan with exactly these sections:
SCOPE
Three lists: IN, OUT, UNCLEAR. If the brief does not mention something
a project like this would usually involve, put it under UNCLEAR.
MILESTONES
Five to eight outcomes, in a sensible order. For each, one sentence
describing how we will know it is reached. No dates, no durations.
ASSUMPTIONS
Everything the brief takes for granted about budget, people, systems,
suppliers, approvals and timing. One line each.
OPEN QUESTIONS
What must be answered before this plan can firm up. For each, suggest
the role most likely to know, but do not invent names.
Rules:

Only use information in the brief and my notes.
Do not estimate effort, cost or time.
Where the brief is vague, say so rather than filling the gap.
Keep it to one page.
BRIEF AND NOTES:
[paste here]

When the brief is a conversation

Sometimes there is no email at all. The brief arrives as a ten-minute chat in a corridor, or a sponsor thinking out loud on a call. In that case, write down what you heard as soon as you can, in your own words, including the bits you are not sure about. Mark anything you are guessing with a question mark. Then feed those notes into the same prompt.

Do not be tempted to tidy your notes first. The hesitations are useful. "Sam said end of year, I think, but may have meant financial year" is exactly the kind of line that should end up under OPEN QUESTIONS, and a polished version would have quietly settled it for you.

Then argue with it

The output is a draft for you to attack, not a plan to send. Read it with the brief open. Anything in SCOPE that you cannot trace back to the brief or your notes, delete. Anything missing from ASSUMPTIONS that you know from experience, add. The model does not know that the last three portal projects stalled at security review. You do.

Once you have edited it, one more prompt is worth running, because it catches the gaps you have become blind to.

Here is my first plan. Read it as a sceptical sponsor would.
List the five questions you would ask in the first meeting that this
plan cannot yet answer. Do not rewrite the plan. Just the questions.

Then take the open questions to the people who can answer them. That conversation is the actual planning. Everything before it was preparation.