Copy text froma screen share
Someone shares their screen on Zoom, Teams, or Meet, and the one thing you need, an ID, a config value, the acceptance criteria on their slide, is unselectable, because a share is video, not text. Khint gets it out live: open the palette, drag a box over the shared window, and the words are on your clipboard before the presenter moves on. No bot in the call, no recording.
It's their screen,
so nothing is selectable
In your own apps, text is text: you select it and copy it. The moment a colleague hits Share, that stops being true. Their terminal, their Jira board, their spec, all of it arrives on your screen as a video stream your cursor can't touch. So meetings fill with the small rituals around it: "can you drop that in the chat?", a screenshots folder full of slides to retype later, a phone photo of a monitor. Khint's Capturepillar treats the share like any other pixels: drag a box over the region and get the characters back. It's one of Khint's three pillars alongside Agents and Memory, all behind the same palette shortcut.
Get the text without stopping the meeting
The same string off a shared slide, two ways.
- “Sorry, can you go back a slide?”
- Ask the presenter to paste it in chat
- Screenshot it, retype it after the call
- Photograph your monitor with your phone
- Cmd+Shift+K → Extract text
- Drag a box over the shared window
- Text is on your clipboard
- The presenter never has to stop
Mid-call, step by step
From someone else's slide to your clipboard, then your ticket.
Open the palette on top of the call
A dev is sharing an error trace, a client is walking through a requirements table, an architect has a diagram up. Hit Cmd+Shift+Kand Khint's palette opens over the meeting window, whatever app the call is in. Capturelives there next to your Actions, so there's nothing else to launch mid-call.
Extract text: drag a box over the shared window
Pick Extract text and draw a box around the part of the share you need: the JQL someone pasted into their terminal, the row of IDs in a spreadsheet, the acceptance criteria on a slide. On macOS the selector is the system screenshot crosshair; on Windows, Khint draws its own region selector. Their pixels, your box.
The text is on your clipboard before the slide changes
Khint reads the region and copies the plain text straight to your clipboard, with a one-line overlay preview confirming the grab. No window opens, so you never visibly leave the call. Paste it into your notes, a ticket draft, or the meeting chat, and let the presenter keep going instead of asking them to stop and paste it themselves.
Ask about it when it isn't just text
When the share is a dense dashboard or an architecture diagram, choose Capture & ask instead: the same region gesture opens a multi-turn Vision chat on what you captured, so you can question the screen while the discussion is still live. Ask AI about anything on your screen →
The grab outlives the meeting
Every capture lands in your local history, and the Capture page keeps a Recent captures feed where each grab can be expanded and copied again, so nothing evaporates when the call ends. With a Memory session active, the capture is logged into your session too, and a later Action, say, drafting the follow-up ticket, runs with that context. With Jira or Linear connected on the integrations page, the ticket can be filed from the same palette. Screenshot an error into a bug ticket →
Works with every meeting app, joins none of them
Khint doesn't plug into Zoom, Teams, Meet, Webex, or Slack huddles, and that's the point: it reads what's on yourscreen, so any app that can show you a share is already supported, including the desktop client your company standardised on. There's no bot participant for attendees to notice, no meeting recording, and nothing captured until the moment you drag a box. That also keeps the permission story small: on macOS the region grab is delegated to the system screenshot UI, so Khint never asks for Screen Recording access, only Accessibility to paste the result back. If what you really want from a meeting is structured notes rather than one string off a slide, action-item extraction is the better-fitting workflow; this page is about the moment something on the shared screen has to come home with you.
Common questions
How do I copy text from a Zoom or Teams screen share?
You can't select it, because a screen share is video: the presenter's text reaches you as pixels, not characters. With Khint, open the palette with Cmd+Shift+K during the call, pick Extract text, and drag a box over the part of the shared window you need. Khint reads the region and puts the plain text on your clipboard, with a small overlay confirming what it grabbed. It works the same in Zoom, Teams, Meet, Webex, or a Slack huddle, because Khint reads your screen rather than integrating with any particular meeting app.
Does Khint join or record my meetings?
No. Khint is not a meeting bot: it never joins a call, never records audio or video, and captures nothing on its own. A capture happens only at the moment you trigger it, and only for the region you drag. On macOS it doesn't even ask for Screen Recording permission: the region selection is delegated to screencaptureui, the system screenshot UI, so Khint's own process never calls a screen-capture API. The only macOS permission it needs is Accessibility, to paste text back for you.
What if they're sharing a diagram or a dashboard, not plain text?
Then a raw text grab isn't enough, and the palette's other capture row, Capture & ask, is the better move: it opens a multi-turn Vision chat on the region you selected. Ask what the architecture diagram implies for your feature, or which figure in the dashboard is the one being discussed, and follow up in the same chat. For pure text, though, Extract text is faster: one gesture and it's on your clipboard.
Where does the capture go afterwards?
The image of the region is sent to Khint's vision service to read the text, and the recognised words come back to your clipboard and your local history: the Capture page keeps a Recent captures feed where each grab can be reopened and copied again. History bodies stay in a local database on your machine and are never synced to Khint's servers. If a Memory work session is active, the capture is also logged as an entry there, so it becomes part of the context your later AI actions can draw on. Captures have their own daily allowance, separate from AI actions.
Never ask "can you paste that in the chat?" again
Free with 10 AI actions and 5 captures per day. No credit card. Capture lives in the palette, one shortcut away.