Three years ago, I needed 90 minutes to prep a hard meeting and still walked in unsure of the angle.
Today the same prep takes 8 minutes, and the meeting goes differently.
The change is not longer hours or a better deck template. It is one system that uses AI to do in 8 minutes the part of prep I was always bad at: compressing scattered context into a single page.
Here is the 5-step format I run before any meeting that matters.
The premise
Most corporate prep fails for the same reason most corporate emails fail. The work that matters is buried under the work that feels like preparing.
Re-reading the deck for the fourth time feels like prep. Skimming the email thread one more time feels like prep. Neither of those tells you what the counterpart will actually push back with on Tuesday at 9 AM.
The 8-minute system flips the order. Compress the noise first. Decide what to do with the compressed version second.
What used to take 90 minutes of re-reading now takes 4 minutes of model output and 3 minutes of trimming. The compounding effect across a week of meetings is bigger than any productivity tool I have tried.
Step 1. Pull every input into one folder before you open the model
Most people start with the wrong prompt. They open Claude or GPT and type “help me prep for tomorrow’s meeting.”
The model has no idea what tomorrow’s meeting is. Garbage in, generic out.
Spend 60 seconds dragging every relevant artifact into one folder. The calendar invite. The last deck the counterpart sent. The email thread that triggered the meeting. Slack and Teams threads where the topic came up. Past notes if you have them.
The first time you do this you will realize how scattered the context actually is. That is why most prep is bad. You were not preparing. You were collecting.
Action: never start a prep prompt without the folder ready. If the folder is empty, the meeting is not ready to be prepped.
Step 2. Run the context-collapse prompt
Same prompt, every meeting:
“Read everything in this folder. In 5 bullets, tell me: (1) who is in the room, (2) what decision is being asked, (3) what the counterpart’s last position was, (4) what has changed since last contact, (5) what would make this meeting a waste of time.”
The five bullets force the model to compress the noise into something you can scan in 30 seconds. Especially bullet 5. Most prep skips the “what would make this a waste” question, and that is the question that surfaces the hidden traps.
The reason the prompt does not change across meetings is that you want the output shape to stay constant. After a few weeks, you start scanning the 5 bullets faster than you scan an email summary. That speed is the whole point.
Action: paste the same prompt every meeting. Do not rewrite it. The repetition is what makes the output comparable.
Step 3. Name the counterpart’s likely move before you walk in
Now one follow-up:
“Given the position the counterpart held last time, what is the most likely move they make in this meeting? What is the second most likely? What would surprise me?”
This is the step that beats most senior decks I have sat through.
Decks tell you what you will present. Very few tell you what the other side will probably push back with. The other side is the part you have to know cold.
Across years sitting in rooms where the other side ran the prep better than I did, this is the question I wish I had asked earlier. Now it costs nothing.
Action: do not walk in without naming the counterpart’s likely move out loud. If you cannot name it, you are walking in blind.— Franco