Notetaker
Listens to meetings and produces written meeting intelligence.
- Records and transcribes the meeting
- Identifies speakers
- Creates a summary, decisions, and action items
- Keeps the transcript available after the meeting
Back to selected work
Wispr hears the meeting. What if it also understood the screen?
Today, Notetaker listens to a meeting and creates notes. But if someone shares a product screen and says “change this banner,” the note keeps the words—not which banner they meant. I am proposing an optional way to save that missing visual context.
- Wispr today
- Meeting audio → transcript and notes
- What is missing
- The exact screen behind a comment
- My proposal
- Optional screen context for key decisions
- Main result
- A clear visual action brief
The proposal in one lineWords + exact screen → a task everyone understands
01 · Today
What Wispr currently does.
Wispr has two separate products. Notetaker captures meetings. The Fn Flow bar dictates into whichever text field is active. This proposal changes Notetaker only; the Fn dictation feature is not part of it.
What the final note contains
In my live test, Wispr returned notes, a summary, a transcript, and action-item fields. It did not return screenshots, pointer locations, or screen-state references.
That visual gap—not the Fn bar—is the problem this concept addresses.
02 · What gets lost
The transcript hears “this.” It cannot see what “this” means.
What someone says in the meeting
“Make this background light yellow, bold the timer, and keep the same 20-minute rule.”
What happens next today
Someone reopens the product, finds the screen, takes a screenshot, rewrites the decision, and explains it again to a designer, engineer, or agent.
03 · Proposal
Add one optional feature: Screen Context.
Before a screen-based review, Wispr asks whether the presenter wants to share one screen, window, or tab. If they agree, Wispr can connect important comments to the exact thing being shown.
Audio → transcript → summary and actions
Audio + selected screen + pointer → clear visual action
User chooses one screen, window, or browser tab
Wispr saves only important visual decision moments
Each moment links the words, screenshot, and pointer
The final action shows the exact component, change, and owner
04 · How it works
Four steps. Normal Notetaker stays intact.
Screen Context is an add-on for meetings where the screen matters. Audio-only meetings can continue exactly as they do today.
Keep normal Notetaker on
Wispr continues recording, transcribing, and summarising the meeting exactly as it does today.
Optionally share one screen
The presenter chooses one screen, app window, or browser tab for this meeting only.
Save important visual moments
When a decision is made, Wispr links the quote to the visible screen and pointer location.
Review a visual action brief
After the meeting, the user checks the screenshot, requested change, owner, and source quote before sharing it.
05 · Result
The first output should be one clear visual action brief.
The goal is not to collect more meeting content. It is to remove the manual work of explaining the same visual decision again.
Primary proposed output
Visual action brief
- Screenshot of the exact screen state
- Highlighted component or pointer location
- The decision in plain language
- Owner and expected result
- Link to the original transcript moment
Example actionChange the highlighted grace-period banner to light yellow, bold “20 minutes,” and keep the backend timer unchanged.
Possible later extensions—not required for version oneShort visual recap Reviewed agent package
Replay only the important screen moments.
Send the approved brief to Codex or Claude.
Interactive conceptSee the proposed permission, capture, and visual brief flow.
06 · Safety
The user must always know what Wispr can see.
Screen access is more sensitive than meeting audio. The feature only works if permission, visibility, deletion, and review are designed into the main experience.
Permission is requested for each meeting
Only one user-selected source is visible
Screen Context can pause while transcription continues
Notifications and sensitive fields are hidden
Individual screenshots can be removed
Nothing is sent to an agent without approval
Clear boundary
This is an independent proposal, not a current Wispr feature. It should begin with one selected source, key screenshots, and a human-reviewed brief—never continuous surveillance or automatic execution.
The recommendation