PromptPack4 - Status Report

<aside> 🎬

Vibhor Chandel · AI for Professionals

From 20 years of running projects and teams

New videos every Wednesday

YouTube · LinkedIn · Substack

</aside>

These are the prompts from my video on writing a weekly status report with ChatGPT. You get a fill-in template for each part of the BOSS Report, the whole main prompt in one piece, the exact prompts I used for the Fernhill office move in the video, and the sample updates so you can run it yourself.

<aside> 📥

Downloads

PromptPack4 - Status Report (PDF)

Fernhill office move: the five updates from the video, with the two off-the-record lines still in (zip)

</aside>

In this pack

<aside> 📌

Before you start

Check your company's AI rules before you paste work updates into any AI tool. Some companies say no to this, and that's their call.
Start a new chat for each report, and send the prompts in order in that one chat.
Replace everything in [square brackets] with your own details. </aside>

🎯 Step 1 · B: Bottom line

<aside> 🎯

The bottom line is the first line your boss reads. It says what you need, from whom, and by when. Don't let ChatGPT guess it: ask it to interview you first.

Prompt 1: the interview prompt (send it on its own)

Before you write anything, ask me three questions, one at a time: who is reading this report, what I need from them, and by when. Don't draft the report until I've answered all three.

Your three answers (send each one when ChatGPT asks)

Who: [Names and roles of the people reading it]. They read it [where and how, for example on their phones between meetings].

What: [What you need from them: a decision, an approval, help]. [The options or details, in one or two sentences.]

By when: [Day, date], so [what has to happen after their answer].

<aside> 💡

Quick tip. If you don't need anything from your boss this week, say so: answer "Nothing needed this week." If you leave it blank, ChatGPT will invent homework for your boss just to sound helpful.

</aside>

🚦 Step 2 · O: On track or not

<aside> 🚦

Bosses have been burned by green reports that weren't green, so a colour on its own means nothing to them. Tell ChatGPT what each colour means, and ask for the reason in the same sentence. This line goes into the main prompt.

On track or not: green, amber or red, with the reason in the same sentence. Green means on track, no help needed. Amber means at risk, and we have a plan. Red means off track, and I need help to recover.

<aside> 💡

Quick tip. Amber isn't bad news. With this definition, amber means you spotted the problem early and you already have a plan.

</aside>

📈 Step 3 · S: Since last week

<aside> 📈

Your boss doesn't need proof you were busy. They want to know what changed because of your work. This line goes into the main prompt.

Since last week: results, not meetings. Three bullets at most.

Before you paste your updates:

Label each paste at the top: what it is, who it's from, and the date. For example: [Email, Tom Okafor, Wed Nov 18]. That way ChatGPT knows who said what, and when.
Put this guardrail right above your updates (it's already in the main prompt): "Use only the updates I paste below. If something isn't in them, write "Not in the updates" instead of guessing."
Delete anything someone told you off the record before you paste. ChatGPT doesn't understand office politics.
Using a personal ChatGPT account? Click your name, then Settings, then Data controls, and turn off "Improve the model for everyone".

<aside> 💡

Quick tip. Search your updates for words like "between us", "off the record" or "not for the report". It takes about ten seconds.

</aside>

⚠️ Step 4 · S: Snags

<aside> ⚠️

Keep issues and risks apart. An issue has already happened and needs action now. A risk might happen, so it needs a trigger to watch and a plan. This line goes into the main prompt.

Snags: issues and risks in two separate lists. An issue has already happened: what it is, the impact, the owner and the next step. A risk might happen: what it is, what would trigger it, the owner and what we're doing about it. If there's no owner, write "No owner yet". Don't pick one.

<aside> 💡

Quick tip. Keep "Don't pick one." If a problem has no owner in your updates, ChatGPT will happily give it to someone from your emails. Putting a name next to a problem that person never agreed to own is a fast way to lose their trust.

</aside>

🧩 The main prompt, all in one

<aside> 🧩

The video builds the main prompt one letter at a time. Here it is in one piece (prompt 2). Send it after your three answers, with your labelled updates pasted under "My updates:".

Now write my weekly status report for [project], using my three answers above.

Use only the updates I paste below. If something isn't in them, write "Not in the updates" instead of guessing.

Write it in this order:

Subject line: project, status colour, and the ask with its date.
Bottom line: what I need, from whom, and by when.
On track or not: green, amber or red, with the reason in the same sentence. Green means on track, no help needed. Amber means at risk, and we have a plan. Red means off track, and I need help to recover.
Since last week: results, not meetings. Three bullets at most.
Snags: issues and risks in two separate lists. An issue has already happened: what it is, the impact, the owner and the next step. A risk might happen: what it is, what would trigger it, the owner and what we're doing about it. If there's no owner, write "No owner yet". Don't pick one.
Next week: three bullets at most.

Keep it under 200 words, so it fits on one phone screen.

My updates:

[Paste your labelled updates here]

</aside>

🔍 Step 5: The mistake prompt

<aside> 🔍

Don't hit send yet. Ask ChatGPT to check its own draft against your updates, line by line, then fix what it found. Both prompts work on any report as they are.

Prompt 3: the mistake prompt

Now check your draft against my updates, line by line. Treat my three answers above as part of my updates. Make a table with three columns: the line in the report, the exact words in my updates that support it, and a verdict (supported, changed or not found). Flag every date, number, name and decision that doesn't match my updates word for word, anything written as decided that was only suggested, and anything I asked to keep out. Don't rewrite the report yet.

Prompt 4: the fix prompt

Fix the report using your table. Correct every line marked changed or not found so it matches my updates, or take it out. Keep the same order, and keep it under 200 words.

Then read the fixed report once through your boss's eyes. If something that matters is still missing, tell ChatGPT exactly what to add.

Prompt 5: your own read

You missed one thing that matters for the decision. [The missing fact, with its numbers]. Add that to the bottom line. Keep everything else the same and keep it under 200 words.

<aside> 💡

Quick tip. The fix prompt corrects the lines in the table. Anything ChatGPT flagged below the table can still be left out, even when it suggests adding it back. That's why your own read comes last.

</aside>

📱 Step 6: Fit it to the reader

<aside> 📱

Once the facts are checked, reshape them for how your boss reads: a short chat message, or one slide for a meeting.

Prompt 6: the short versions

Using the fixed report above, give me two versions:

A three-line Teams message for [name]: the ask, the status and its reason, and the date.
One slide: a title that states the ask, then status, what changed, issues and risks. Don't add anything that isn't in the fixed report.

<aside> 💡

Quick tip. Check first, and shorten second. A short version made from an unchecked report just carries the mistakes into fewer words.

</aside>