# STEP 7: DESIGN YOUR DAILY BRIEF **Goal:** Decide together what should be waiting for them every morning, then run it live once so they see the real thing before anything gets automated. **Time:** 15 minutes. --- ## THE RULE Do not ask "what would you like in your daily brief?" It is a terrible question. Nobody knows. They will say "a summary of my day" and you will build something generic that they stop reading in a week. Instead: **propose, then let them cut.** You have their calendar, their inbox, their document and the thing they said they dread. You are better placed to design this than they are. Design it, show it, let them edit. --- ## BUILD THE PROPOSAL FROM WHAT THEY DREADED Go back to the interview answer about the part of their week they dread. That is the brief's real job. Everything else is decoration. If they said scheduling is the misery, the brief leads with the double bookings and the gaps and who still has not confirmed. If they said following up, it leads with everyone waiting on them and how long they have waited. If they said losing track of deals, it leads with what moved and what went quiet. Say that out loud when you propose it: "You told me the thing you hate is X, so this brief attacks X first and everything else is secondary." --- ## PROPOSE THE SHAPE Give them four or five sections with a one line description each, ordered by what matters to them. Something in this shape, adapted heavily: > **Your morning brief, first draft:** > > 1. **The one thing.** The single most important thing today, and why. > 2. **Today, hour by hour.** Every meeting with a prep note, who they are, what > happened last time, what to bring. > 3. **Waiting on you.** People who need an answer, longest wait first. > 4. **Went quiet.** Things you were expecting that have not come back. > 5. **Against your goal.** One line on whether today moves the ninety day thing > you told me about, or does not. > > Cut anything you would not read at seven in the morning. Then shut up and let them cut. Encourage cutting. A brief they read every day beats a complete one they skim once. --- ## RUN IT LIVE, RIGHT NOW This is the step people skip and it is the most important one in the session. Build the brief prompt, then run it immediately against their real data and show them today's actual brief. The brief itself, built from their real week. Then ask the only question that matters: **"Would you have read all of that?"** --- ## THE LOOP, AND THE RULE THAT GOVERNS IT **Every single edit produces a new full example. No exceptions.** This is the rule people break, including you if you are not careful. When they ask for a change, the tempting move is to describe what you changed and ask if that sounds right. Do not do that. They cannot approve a description. They are about to put this on a timer and stop watching it, so the only thing worth their judgment is the actual output. So the loop is: 1. They ask for a change. 2. You change the prompt. 3. You **run it again against their real data and show the whole brief.** Not the changed section. The whole thing, top to bottom, as it would arrive. 4. You ask again whether they would read it. Repeat until yes. Two or three rounds is normal and a sign it is working. Five is fine. Never cut the loop short because it feels repetitive to you. It is not repetitive to them, it is the only look they get. **Never show a partial brief.** A section in isolation always reads better than it does inside the whole. The order and the length are most of what makes a brief readable or not, and neither is visible in a fragment. --- ## THE APPROVAL Do not slide from the loop into the next step. Stop and get a clear yes. Show the final version one last time, whole, and say plainly what they are approving and why it matters: > This is the brief. This exact thing arrives every morning without me asking you > again. Say yes and I will lock it in, or tell me what to change and I will > rebuild it and show you again. **What counts as approval:** a clear yes to that question. **What does not count:** "looks good", "sure", silence, or moving on to another topic. If you get any of those, ask once more, directly: "Would you be happy to open exactly that tomorrow morning?" If they hesitate at all, something in it is wrong and they cannot name it yet. Help them find it. Ask which section they skipped when they read it. That is almost always the answer. **Log the approval.** Write down that they approved it and what the approved version was. Step 9 will check for this before it schedules anything. **If their connections are off or they are on the free plan:** ask them to paste in tomorrow's calendar or a few lines about what is on their plate, then produce the brief from that. Same loop, same question. What you are proving is the shape, and the shape works either way. --- ## SAVE THE PROMPT Only after approval. Write out the final brief prompt as a clean block of text they can copy, and have them add it to the Project knowledge alongside the rest. Name it plainly. Save the approved example output next to it. Two things in the Project: the prompt that generates the brief, and one real example of what it produced. The example is what they will compare against in a month when they think something has drifted, and it is what Step 9 checks against after the first scheduled run. --- ## PROOF GATE All three: 1. They have seen the complete brief, built from real information, in its final form after the last edit was made. 2. They have given a clear yes to the question of whether that exact thing should arrive every morning. 3. The prompt and the approved example are both saved in the Project. If an edit was made after the last time you showed them the whole brief, you have not met this gate. Run it again. --- ## NEXT Show the progress line, then open `https://claudefoundersguide.com/setup/07-brief-delivery.md`.