How to Brief an AI Workforce When You Have a Goal, Not a Task List

Six · 6 min read · October 14, 2025

How to Brief an AI Workforce When You Have a Goal, Not a Task List

Say you run operations, and on Monday the CEO tells you Thursday's leadership meeting needs a better report on the customer renewal program. The current report lists the renewal numbers every week, and nobody in the room seems sure what to do with them.

You can see what a better report would do. What you don't have is a list of tasks, and you won't have one before Thursday unless you stop and write it yourself.

Most work starts like this. Someone can see the outcome and hasn't turned it into steps yet. If the only way to get help is to hand over a finished task list, you do the planning alone before anyone else touches the work. A brief is a better first move. It records the decision ahead, the material you already have, the deadline and the questions only a person can settle.

Start with the decision the report supports

"Prepare a better leadership report" sounds clear until you ask what the report is meant to change. Is the leadership team deciding whether to keep funding the program? Choosing between it and two other priorities? Looking for risks that need an owner?

Each of those is a different assignment. A funding decision needs what the program costs and what it has brought back. A list of risks needs a name next to each one. The decision even changes who should read the draft first.

So write the decision in one sentence before you name the deliverable:

At Thursday's meeting, we need to decide whether to keep investing in the renewal program, change its approach, or stop it.

Now the report has a job. It has to make that choice easier. It doesn't need to recount every renewal conversation from the last quarter.

The GOV.UK Service Manual asks teams building public services to write user needs that "focus on the user's problem rather than possible solutions." The user stories that describe specific features are written to meet those needs. Government services are a long way from a leadership meeting, but the habit carries over. The CEO asked for a report. The problem behind it is a decision about the renewal program.

Say what you already have

You don't need a tidy research library before you can hand off the work. You do need to say what exists and what each piece is good for.

For the renewal report, that's the last two weekly reports, renewal data from the finance owner, notes from three recent customer calls and the current program budget. There's also a gap: nobody has agreed on why the renewal rate changed.

Put that gap in the brief. An unknown isn't an empty field for someone to fill with a plausible answer. Someone may need to supply it, or the work can go ahead on an assumption that's written down. Sometimes the question matters enough to hold the work until its owner decides. Which of those is right depends on what it costs to be wrong. If leadership is deciding whether to stop the program, a confident guess about why the rate changed could end something that's working.

Ask only the questions that change the work

Once the decision and the material are written down, most briefs need only a few more answers.

Ask Why it changes the work
What decision must this support? It determines the recommendation and the evidence needed.
Who will read or approve it? It sets the audience, the tone and who reviews it.
What material is already trusted? It stops anyone hunting for what you already have, and shows what's missing.
What is the deadline? It decides how much can be done.
What is off limits? It keeps the work inside the authority and context you intended.

Don't turn those five into a ritual. If the CEO's request already says who will read the report, carry that forward and don't ask again. If someone doesn't know an answer yet, "not decided yet" is an answer, and it belongs in the brief.

GOV.UK's guidance on question pages, written for people designing online forms, says much the same: know why you're asking each question, "only ask for a piece of information once within a single journey," and let people answer "I do not know" when that's a valid response. A brief isn't a form, but the CEO will notice if you ask who's reading the report after they've already told you.

For the renewal report, two of those answers aren't settled. The commercial owner hasn't said whether the recommendation can propose a budget change. And the finance data may not cover the same dates as the weekly reports, so someone has to check it before the two are compared.

The brief for Thursday

Put together, the brief is short:

Goal: Give the leadership team a clear basis for deciding the next quarter of the renewal program.

Decision: Continue the program as it is, change the offer or the outreach, or pause further spending.

What we have: The last two weekly reports, the program budget, renewal data from finance and notes from three recent customer calls.

Deadline: A draft by Wednesday afternoon for the Thursday meeting.

Still open: Whether the recommendation can propose a budget change, which the commercial owner decides. Whether the finance data covers the same dates as the weekly reports. Why the renewal rate changed.

What good looks like: A short report that keeps what was observed apart from what it might mean, names the decision, says what evidence is missing and gives the leadership team options they can act on.

There's no list of twelve tasks in it, and the work can still be organized from it. The weekly reports can be compared, and the finance data gets its date check before anyone draws a trend from it. The budget question goes back to the commercial owner instead of being answered by guesswork.

Read it back before work starts

Before anyone starts, read the brief as the person who will make the decision. Would the CEO recognize what's being decided? Do the inputs look right? Would the deadline, the audience or anything you left out surprise them?

GOV.UK's check answers pattern does the same job for online forms. It shows people what they've entered before they submit, which gives them "a second chance to notice and correct errors before submitting data."

Say you send the brief over and the CEO replies that stopping the program isn't on the table for next quarter. The real question is whether to keep it as it is or change it. That one reply changes the work. The report no longer has to make a case for stopping, and the budget question gets more important, because changing the program probably means moving money. Better to hear that on Monday than after the draft is written on Wednesday.

Handing the brief to an autonomous workforce

Say you hand the renewal brief to Supanova on Monday. You don't write anything new: a Supanova project starts with a brief, and your atoms do the work from it. This one is the page the CEO already read, now with two options instead of three and Wednesday's deadline still on it.

The open questions stay with their owners after the handoff. Whether a person or an atom drafts the report, the budget question is still the commercial owner's, and the brief says so. When the draft is ready, read it the way the CEO will on Thursday. Does it make the choice between keeping the program and changing it easier? If it doesn't, compare it with the brief. Either the draft missed something the brief asked for, or the brief never asked. If it's the second, add that line to the next brief.

You don't need a task list to start. Pick one request on your desk that arrived as an outcome. Write down the decision it supports, what you already have and what's still open, and show it to the person who will decide before any work begins.

When you're ready to hand that brief to a workforce, start at supanova.team.

Sources

autonomous-workforceproject-briefsdelegationgetting-started