Smart Meeting Prep
Predict the five topics that matter before every one-on-one
- Difficulty
- Starter
- Time to result
- ~days to results
- Steps
- 5
- Confidence
- 98%
Smart Meeting Prep uses connected workplace history to predict what another person is likely to raise in an upcoming meeting. The inputs are the attendee's identity, prior communications, shared documents, project records, one-on-one notes, and calendar context. An AI assistant searches those sources and returns five likely discussion topics. The leader then validates the suggestions, removes irrelevant items, and converts the strongest topics into an agenda. This replaces vague preparation based on memory with an evidence-backed preview of the other person's probable priorities. Its effectiveness depends on connector coverage: if important work happens in Slack, Asana, HubSpot, or another system that is not connected, the resulting agenda may omit critical issues.
Origin
Adapted from a Satya Nadella prompt and demonstrated on Marketing Against The Grain with connected workplace data.
Core principles
- 01Prepare from shared evidence, not memory alone
- 02Predict the other person's priorities before setting the agenda
- 03Connect AI to the systems where work actually happens
- 04Limit the output to a small set of high-value topics
How to run it
- 1
Name the meeting counterpart
Specify the person you will meet so the assistant can isolate the relevant shared history.
Pro tip Include the meeting type, such as a one-on-one or project review.
- 2
Connect the work context
Give the assistant access to the communication, document, calendar, and project systems where your collaboration is recorded.
Pro tip Explicitly name the connectors that contain the best evidence.
Watch out Missing systems can create a falsely incomplete agenda.
- 3
Prioritize shared records
Direct the assistant toward shared project documents and one-on-one notes rather than asking it to search indiscriminately.
Pro tip Add a relevant time window when the relationship has a long history.
- 4
Predict five topics
Ask for five subjects the other person will likely want to discuss. A fixed number forces prioritization.
Pro tip Request a short evidence note for each topic if you need to validate the result quickly.
- 5
Build the agenda
Check the proposed topics against current priorities and turn the credible items into a focused meeting agenda.
Pro tip Share the agenda before the meeting when alignment would save time.
Watch out Do not treat speculative topics as confirmed concerns.
In the wild
A marketing leader connects email, Drive, calendar, and Asana before meeting a colleague. The assistant identifies an upcoming event, budget implications, AI strategy, campaign integration, and internal adoption as the five most likely topics. The leader validates four of them and sends a concise agenda in advance.
→ Both participants enter the meeting prepared to address the highest-value issues.
Common mistakes
Searching the wrong communication system
If most collaboration happens in Slack but only email is connected, the assistant will reason from an incomplete record.
Accepting every predicted topic
The output is a preparation aid, not proof of what the other person intends to discuss; validate each item before using it.
Is it for you?
Best for
Leaders and managers who regularly meet colleagues across several active projects.
Not ideal for
Sensitive meetings where the connected systems contain little context or should not be queried by an AI assistant.
From the transcript
“What are five topics they'll likely want to discuss?”
“This way you can actually preempt what are the most important things to talk about.”
“So anyone you have a meeting with, you can run this prompt and you can get all of the details you need to make sure…”
From the episode
I Stole Microsoft CEO’s 4 Most Powerful AI Prompts