AI Meeting Notes: Better Prompts and Output Review
The same meeting notes can serve different purposes. This practical, hypothetical example shows how to write prompts that specify the purpose, audience, source material, and output format, and how to review the results.
When asking AI to organize meeting notes, start by explaining what you expect “organize” to mean. The information needed to help someone understand a meeting they missed differs from what a team leader needs to assign follow-up work. A useful starting point is to specify the purpose, intended reader, source material, and desired format.
If you are preparing next year’s AI training toward the end of the year, use practical exercises to check whether employees can explain these task requirements independently.
The Same Meeting Notes Can Require Different Outputs
Imagine that a customer support team has just finished a meeting about revising customer guidance materials and needs to create a list of follow-up tasks. The following fictional notes illustrate the scenario:
The customer support team will compile a list of recurring customer questions.
The product team will verify the feature descriptions to be included in the guidance materials.
The release date for the materials and the deadlines for individual tasks have not yet been set.
The notes do not name the individuals responsible for each task.
A brief summary may be enough to share with a colleague. But if a team leader needs to assign the work, the responsible teams and unresolved details must be clearly visible. Defining the output you need first makes it easier to write a specific request for AI.
Here, a prompt means the instructions and reference information you provide to AI.
Add Task Requirements to a Simple Request
You might begin with a request like this:
Summarize the meeting notes below.
To reflect the purpose of this task, you could revise it as follows:
Based on the meeting notes below, create a list that a team leader can use to assign follow-up tasks.
The reader is the customer support team leader, who did not attend the meeting.
Use only the information in the provided notes.
Organize the output in this order: “Task / Responsible Team / Task Owner / Deadline / Items to Confirm.”
Mark any information that is missing or has not yet been decided as “Needs confirmation,” and add follow-up questions at the end.
Each addition serves a specific purpose:
- Purpose: Help the team leader assign follow-up work.
- Audience: The customer support team leader, who did not attend the meeting.
- Source material: Use the provided meeting notes as the basis for the output.
- Format: Clearly separate tasks, responsibilities, and unresolved details.
OpenAI’s official prompting guidance also recommends being specific about the context, desired outcome, and format, and providing an example of the output when helpful. However, no wording guarantees the same results across every tool and task.
Review the Output Against Task Requirements
The following illustrates the expected output for this scenario. It was written to demonstrate the review criteria; it is not the result of running a particular AI tool.
Task 1
- Task: Compile a list of recurring customer questions.
- Responsible team: Customer Support
- Task owner: Needs confirmation
- Deadline: Needs confirmation
- Items to confirm: Who will complete the task and by when.
Task 2
- Task: Verify the feature descriptions to be included in the guidance materials.
- Responsible team: Product
- Task owner: Needs confirmation
- Deadline: Needs confirmation
- Items to confirm: Who will complete the task and by when.
Follow-up questions: Who will own each task, and what are the deadlines? When should the guidance materials be released?
Review actual AI-generated output alongside the original notes. Check that no tasks have been omitted, that responsibilities remain assigned to the correct teams, and that undecided deadlines have not been presented as confirmed.
Even if your prompt instructs AI to use “Needs confirmation,” keep the review step. Whether AI followed that instruction is itself something to verify.
Assess Both Prompting and Review During Training
A shared template can help employees get started with recurring tasks. When the purpose and source material vary, employees also need practice adapting that template. When designing training exercises, ask participants to document what they changed and why, alongside their prompts and final deliverables.
An assessment can help establish your organization’s current capability level. Codepresso’s AI Fluent evaluates both the process of collaborating with AI and the resulting deliverables through practical, role-specific missions. Its assessment covers problem definition, prompt design, and output verification. If you are considering it alongside SkillCamp, Codepresso’s in-person training program, start the consultation by identifying the target roles and the tasks employees need to practice.
The meeting notes in this article are illustrative examples. They are neither actual AI Fluent assessment questions nor evidence of training outcomes.
Discuss AI Competency Assessment and Training for Your Organization
References
Written by Codepresso’s Global GTM & Partnerships (GGP) Division
