Black Friday33% off all plans See plans Enter it at checkout

Screen Recording

Screen recording captures your screen straight onto the timeline. The desktop app also tracks where you click, and the take lands with a Smart Zoom camera that pushes in on your clicks and glides between them, so it looks like an edited product demo without any keyframing. On a Mac, the editor also draws a smooth cursor over the recording. Optional AI understanding goes further: it transcribes the take, writes a title, chapters and a summary, lays a storyboard on the board, and lets your terminal agents read the take.

Start a recording

Click Record at the end of the timeline toolbar (tooltip “Record your screen into a new clip”). In the desktop app a floating Screen Recording panel opens. It stays on top of every window and out of the recording, so you can move it anywhere on screen:

  • Capture — Display (the full screen), Window (you pick the window each time) or Area (click Select region… and draw a region). A region you just drew can be named and saved as a preset with Save, so a common capture area is one click away next time. The × on a preset deletes it. Device capture is marked “soon”.
  • Motion — how the Smart Zoom camera moves: Studio (the default), Calm, Precise, Lively, Follow or No zoom. A small preview under the row plays the picked motion on a made-up screen. See Pick the motion.
  • Audio & camera — System audio (on by default) and Microphone (off by default). With both on, the microphone is mixed into the system audio. Hide app while recording works with full-screen capture only: the app window hides the moment recording starts and comes back when you stop, so it never shows up in the take. Camera overlay is marked “Coming soon”.
  • AI understanding — three optional switches, covered below: AI transcript & understanding, Auto chapters & summary (needs the transcript switch) and Create storyboard.

The panel remembers your choices for next time.

Click Start recording. A small control bar appears at the top right of the screen and counts down (“Starting in 3…”) before capture begins; during the countdown its button reads Cancel. While recording, the bar shows the elapsed time with Pause / Resume and Stop, and it is kept out of the capture. An Area capture also draws a glowing border just outside the recorded region, so you can see exactly what is captured.

The toolbar’s Record button turns into a red Stop button with the elapsed time. Click it, the bar’s Stop, or the operating system’s “Stop sharing” control to finish. The bar shows Processing… while the take uploads.

On macOS the app needs two permissions:

  • Screen Recording — without it macOS only lets the app capture its own windows. The app offers Open Settings, Record app only or Cancel. After you turn it on in System Settings ▸ Privacy & Security ▸ Screen Recording, quit and reopen the app.
  • Input Monitoring — the panel reminds you: “On macOS, enable Input Monitoring for this app so your clicks drive auto-zoom.” It lets the recorder see your clicks, which Smart Zoom follows. Without it the recording still works, but it has no automatic zooms.

Pick the motion

The Motion row decides how the camera moves over the take, and on a Mac also how the drawn cursor looks:

MotionCameraCursor
StudioA smooth spring zoom toward your clicks. The default.Trails your hand a little, with a small press on each click.
CalmSlower and gentler, zooms in less; the clicked spot stays in place.Glides with no lag and fades when you stop moving.
PreciseA closer zoom that puts what you click near the middle.Follows your real path, minus the shake, with a ring on each click.
LivelyA fast, punchy zoom that goes in further.Tight, and tilts as it moves.
FollowWhile zoomed in, follows your cursor, not only your clicks.No lag.
No zoomNo camera: the take lands full frame. You can still add zooms later.The Studio look.

You can change the camera’s feel after recording; see Smart Zoom.

In the web version

In a desktop browser, Record opens a small popover instead of the floating panel: Capture area (Full screen or Window), Audio (System audio, Microphone) and Start recording, and the browser asks which screen or window to share. Click tracking, Area capture, the Motion row and the AI understanding options are desktop-app features, so a browser take arrives without automatic zooms. You can still add zooms by hand; see Smart Zoom on any video. Record is hidden on phones and tablets.

The drawn cursor

On macOS, Display and Area recordings use the app’s own recorder. It records at your screen’s full Retina sharpness, at up to 60 fps, and leaves the cursor out of the picture. The editor then draws the cursor back over the recording:

  • It glides instead of jittering, and lands exactly on every click with a small press.
  • It takes the real cursor’s shape: a hand over links, an I-beam over text, and apps’ own cursors.
  • It gets a touch of motion blur when it moves fast.
  • Its look follows the Motion you picked before recording.

The cursor is drawn into the recording’s own picture, so mockups, masks, effects, transitions and the camera treat it like a recorded cursor. System audio and your microphone end up in one track, and pause and resume work as usual.

Window capture, and recordings on Windows and Linux, keep the cursor in the picture as it was recorded. If the Mac recorder can’t start, the take falls back to that recorder on its own.

What lands on the timeline

When you stop, the recording is uploaded and added as a video clip named Screen recording after the last clip on the active channel, and the playhead jumps to its start. If the active channel is a voice channel or the camera channel (or none is selected), a new Screen recording channel is created for it. If the capture is larger than the project canvas, the project grows to fit it (up to 3840 px); an Area capture sets the project to exactly the region’s size and aspect ratio.

If the take has clicks and you didn’t pick No zoom, a camera clip named Smart Zoom joins the project’s camera channel, lined up with the recording. Record again and the new camera clip joins the same channel after the existing ones: a project only ever has one camera channel. When the recording is the first thing in the project, Motion blur is also switched on in the render settings (Subtle), so fast zooms look filmed in the render. The preview is unchanged.

The take is also added to the board it was recorded from, as a video card in the Video Editor Generated assets frame, even if you switched boards before stopping. Alongside the clip, the recorder keeps the cursor path and the clicks that Smart Zoom follows.

Smart Zoom

The Smart Zoom camera zooms in shortly before each click, glides from one click to the next and eases back out when you stop clicking. It is a camera move, so anything you put over the recording (callouts, arrows, logos) travels with the zoom instead of sliding out of frame.

Nothing is baked into keyframes: the camera reads the recording’s clicks while it plays. Trim, split, move or slow down the recording, and the zoom stays on your clicks. The camera clip moves and stretches with its recording.

The status line after recording tells you what happened:

  • “Recording added · Smart Zoom (Studio) follows your clicks”, plus “render at 60 fps for the smoothest zooms” when your render frame rate is lower
  • “Recording added · no zoom” (you picked No zoom)
  • “Recording added · no clicks detected for Smart Zoom”
  • “Recording added · enable Input Monitoring for Smart Zoom” (no click track: on macOS, the permission isn’t granted)

Adjust or remove the zoom

Select the Smart Zoom clip. Its zooms show as blocks on the camera layer, and its settings are in the camera panel under Movement ▸ Zoom:

  • Change the Feel to make the whole camera calmer, closer or livelier.
  • Hover the camera layer and click + to add a zoom; drag a block to move it, drag its ends to change its length, and click × to remove it.
  • Pick a zoom and frame it right in the preview: drag to move the view, scroll or pinch to zoom.
  • Back to automatic zooms drops your edits and zooms on the clicks again.
  • Delete the camera clip to drop the zoom entirely.

Smart Zoom covers every control. A layer that should stay locked to the frame through the zoom (a watermark, a fixed caption) can opt out with the camera button on its channel header.

Projects recorded before Smart Zoom keep their Auto-zoom camera clip, with Zoom, Pan X and Pan Y keyframes you can edit in the camera’s Keyframes tab. See Camera Layer.

AI transcript & understanding

Turn on AI transcript & understanding in the setup panel to transcribe the take in the background after you stop. The status line reads “Recording added · transcribing…”, then “Recording transcript ready” (usually within a minute or two), or “Recording transcribed · no speech detected” for a silent take. Transcription needs an OpenAI or WaveSpeed key in Settings.

Once the transcript is ready:

  • The recording clip shows a CC badge, and its panel gains an AI DATA section listing Transcript (word count and language) and Screen recording AI (transcript · events · frames). The ✕ next to an entry removes that data from the layer after a second click on Remove?.
  • Click Voice in the timeline toolbar with the clip selected to lift the speech onto a voice channel as an already-transcribed take; the video is muted. From there you can edit the narration word by word, and Captions on the take builds a timed caption track. See Voice Channels and Transcribe Editor and Captions.

The take also becomes something a terminal agent can reason about. GLEA, GLAI or an external agent connected to the board can read the transcript, the click and cursor track, and individual video frames on demand. Ask it what happened in a recording, what was on screen at a given moment, or what you clicked. Board Chat in the agent dock doesn’t open recordings, so ask a terminal agent.

Auto chapters & summary

Needs AI transcript & understanding. After transcription, an AI pass detects the scene changes in the take, then titles and summarises it using the transcript, your clicks and sample frames:

  • The clip, and its card on the board, is renamed to the generated title, and chapter marks appear along the clip on the timeline. Hover a mark to see the chapter’s title.
  • The recording gets a summary and tags that terminal agents read when you ask about your footage.
  • The status line reads “Recording chapters & summary ready” when it lands. This runs after transcription, so give it a few minutes on long takes.

The pass uses the first chat model your keys unlock (OpenAI, WaveSpeed, Pika, Anthropic or Google). If it fails, the status line reads “Chapter generation failed — ask the agent to regenerate insights.” A terminal agent can re-run the pass, retitle the take, edit the chapters or re-tag it.

Create storyboard

Turn on Create storyboard to lay the take’s detected scenes onto the board as a “Storyboard for …” frame: one “Scene N” frame per scene with its first, central and last frames and a timecode note, plus a green task note that asks an agent to describe each scene (see Scene Frames and Storyboard). It is placed below your existing board content, and scene detection runs in the background so long takes don’t hold up the editor.

Remove a recording

Deleting a recording clip that has AI data asks whether to delete everything. OK also deletes the source video file, its board card and the AI data (transcript, events, frames); Cancel only removes the clip from the timeline. Removing the clip is undoable; deleting the file and data is not.

Where to go next

Was this helpful?

0

Updated

Comments

Leave a Reply