❗️Custom sections are available on the Business plan.
Creating a custom section
Open a meeting note from the Your meetings page
Scroll to the bottom of your notes and click on the ✨Add Section button
Select Create a section
Enter the title and description of your new section
Click Generate
Review the output and click Done
You can have up to 3 sections per meeting. The Gist does NOT count towards this limit. If you’ve reached the limit, you must remove an existing section to add a new one.
Editing a custom section
You can edit your custom section by hovering over the ellipses icon next to the section title and clicking on the edit icon ✏️.
Note: Whenever you edit a custom section, the changes will take effect in the notes in which you made the edit, as well as future notes you apply that section to. Edits will not apply to previous meeting notes.
Best practices
Supernormal uses state-of-the-art AI models to create sections in your notes. Like all AI models, Supernormal has limitations and the quality of your input can have a large effect on the quality of its outputs.
The tips below will help you craft descriptions that produce meaningful results.
🗓️ Ask for information relevant to your meeting
Meeting transcripts serve as the primary source of information for all your sections. When creating and adding new sections to your meeting notes, it’s important for that section to be relevant to the meeting content to capture accurate and useful information.
✍🏻 Be as specific as possible
Include details in your descriptions to get more relevant outputs.
Less effective ❌ | Better ✅ |
Founding team’s background | How the founding team met, their history of working together, and what they did at their last employer |
Design-related feedback | Design-related positive feedback, negative feedback, and ideas |
Potential offsite location | What were the offisite options that were brought up during the meeting and the pros and cons of each? |
🧮 Don’t ask for numeric calculations
Numeric calculations often require precise interpretation, and their accuracy depends on how clearly the numeric value ares presented in the transcript. Furthermore, while Supernormal’s AI models can understand and generate text-based information, it is not designed specifically for tasks involving arithmetic.
Poor descriptions:
Percentage of words spoken by each meeting attendee
Calculate the difference between the Q2 and Q3 user signup metrics
OK descriptions:
AcmeCorp is a potential customer. What is the customer’s budget?
What was the user churn rate in Q3?
🙅🏻 Don’t request information that may promote biases
We encourage you to be mindful of ethical considerations when using custom prompts to extract information from your meetings. AI models learn from vast datasets, which may include biases present in the data.
To avoid perpetuating these biases, please refrain from requesting information that may promote stereotypes, discrimination, or any form of bias. Anchoring the question on who and/or what was actually discussed during the meeting can help minimize bias.
Bad description ❌ | OK description ✅ |
Who should run this project? | Who did the team decide should run this project? |
Why should we hire this candidate? | Does this candidate meet the qualifications we listed for this role? |