# Screen Recordings — Loom-style walkthroughs

_Automation (Workflows)_

Record your screen, microphone, and (optionally) your webcam right inside CRMown. CRMown AI auto-generates chapters, a transcript, and a summary so every recording is searchable and skimmable. Share with a public link, embed in a Knowledge Base article, or attach to a contact / deal / SOS ticket. Distinct from Operations meeting recordings (those are for Zoom/Meet/Teams calls via the Recall.ai notetaker) — Screen Recordings is for product walkthroughs, async updates, bug reproductions, and onboarding videos.

## Steps
1. Open the sidebar and click Tools > Screen Recordings — available on every plan.
2. Click "+ New recording". Pick a mode: "Screen only" (no audio), "Screen + microphone" (most common), or "Screen + mic + webcam" (adds a picture-in-picture face cam burned into the video).
3. For webcam mode, pick a corner for the PIP overlay (top-left / top-right / bottom-left / bottom-right). The PIP is rendered into the video itself — viewers see it on playback.
4. Click Start recording. Your browser will ask which screen, window, or tab to share. To capture system audio (e.g. for app demos with sound), tick the "Share tab audio" / "Share system audio" checkbox in the browser dialog.
5. A 3-second countdown plays, then recording starts. Switch to the tab or window you want to walk through — the recording continues in the background. The CRMown modal stays open with a live timer.
6. Click Stop when you're done (or the browser's "Stop sharing" banner). Recording auto-stops at 3 hours regardless.
7. Review the recording inline (it plays from your browser, not yet uploaded). Add a title (required) and an optional description, then click Save & process.
8. CRMown AI processes in the background — Whisper transcribes, then a chapter / summary / keywords pass runs. For a 5-minute recording this takes 30-60 seconds; long recordings can take 5-10 minutes. Status flips to "Ready" live via Socket.io — no need to refresh.
9. On the detail page, the tabs are Summary (meeting purpose, key takeaways, topics), Transcript (who said what, click any line to jump there, search), Chapters, Ask CRMown AI, and Analytics. The right-hand column lists Action items (with who owns each and when it was said) and Questions asked. Toggle visibility to "Anyone with link" and copy the share link to send externally — viewers get the same Summary / Transcript / Chapters / Ask CRMown AI / Action items / Questions layout (without Analytics or task buttons).
10. Speaker names: CRMown AI separates the voices and names each speaker from the conversation ("thanks, John") and from name labels visible on screen (e.g. a Google Meet tile). To fix a name, open Transcript, click the speaker chip at the top, type the name and press Enter — it updates every line, action item and question.
11. Turn action items into Tasks: tick the items in the Action items panel and click "Add N to Tasks". Each task links back to the recording and moment. Items already added show "In Tasks →".
12. Copy buttons: "Copy summary" (Summary tab) copies purpose, takeaways, topics and action items as plain text; "Copy items" copies just the action items; "Follow-up email" copies a ready-to-send recap email written by CRMown AI.
13. To cut the start or end off a video, open it and click "Trim" (next to Download). Drag the two teal handles on the timeline to the part you want to keep, or play to a spot and click "Set start here" / "Set end here", or type times like 1:05.5 into the Start and End boxes. Click Preview to watch just the kept part, then "Save trim".

## Tips
- Webcam picture-in-picture is burned INTO the video, not overlaid at playback time. Pick the corner before you start — you can't change it after.
- CRMown AI generates 3-8 chapters per recording, action-oriented titles like "Setting up the integration" or "Demoing the form submission flow". Click any chapter to jump there in the player.
- The transcript is searchable. Open the Transcript tab on the detail page, type into the search box, and matching lines stay visible — click any line to seek to that timestamp.
- Share-link viewers can be required to identify themselves (email + name) before playback — toggle "Identify viewers" on the recording detail page. Useful for client-facing walkthroughs where you want to know who watched.
- Analytics tab shows a 10-second-bucket heatmap of where viewers watched, top drop-off points (where they bail), and country / device breakdowns. Click a heatmap bar or drop-off entry to jump the player to that timestamp.
- Tie a recording to a Knowledge Base article: edit the recording and set "Knowledge article slug" — the recording then surfaces inline on the matching Training article (Tools > Training).
- SOS Helper (the floating ? button) cites screen recordings alongside KB articles. If a viewer asks "how do I set up Stripe?" and you've recorded a walkthrough, the AI links to it in its reply.
- Recordings are stored in CRMown R2 by default. The original master upload is never changed: a trim writes a new published version, and "Restore original" (inside the Trim editor) brings back the full recording at any time.
- Trim keeps ONE continuous part — it cuts the start and/or end, not a section from the middle. Re-trimming always starts from the full original, so you can widen a trim back out.
- Trimming runs in the background. Short recordings finish in seconds; long ones can take several minutes when the video has to be re-encoded for a frame-accurate cut. The current version (and any share link) keeps playing until the new cut is ready, and a "Trimmed" badge appears next to the date.
- After a trim, the transcript and chapters are rebuilt for the new length (no re-transcription), and CRMown AI regenerates the summary when AI chapters are enabled. Download and Save to Folder give you the trimmed version. Analytics from views before the trim still refer to the old timeline.
- For long recordings, keep the browser tab in the foreground. macOS and Windows pause MediaRecorder when the screen locks or the tab is fully backgrounded — the recording resumes when the tab regains focus, but you may see a brief gap.
- CRMown AI processing can be disabled per-org under Settings > AI Usage > AI Features ("Screen Recording AI Chapters"). Disabling skips Whisper and the chaptering pass entirely; recordings remain playable but without chapters or transcript search.

## Example
A SaaS founder recording an async product update walks through 4 features in 8 minutes — screen + mic + webcam in the bottom-right. CRMown AI generates 5 chapters ("Intro", "New billing UI", "Calendar improvements", "Mobile updates", "What's next"), a 3-sentence summary, and a searchable transcript. He sets visibility to "Anyone with link", drops the URL in his team Slack, and goes to lunch. Two days later the analytics show 23 of 31 teammates watched, with 78% completion. The drop-off heatmap reveals everyone bails 30 seconds into the calendar section — feedback gold without a single survey.

## Troubleshooting
- **Browser asks for screen share permission but the modal stays empty / nothing to pick** — macOS Privacy & Security blocks screen recording per-app. The browser doesn't have permission to capture the screen yet. Fix: macOS: System Settings > Privacy & Security > Screen & System Audio Recording. Toggle on for your browser (Chrome, Edge, Firefox, Safari). Quit and re-open the browser. Click "+ New recording" again.
- **Webcam permission denied / no face cam in the recording** — You denied webcam access in the browser permission prompt, or the webcam is in use by another app (Zoom, Teams). Fix: Quit other apps using the webcam. In the browser address bar, click the camera icon and reset to "Ask" or "Allow". Re-open the recorder. If you don't need the webcam, switch to "Screen + microphone" mode.
- **An older recording has no speaker names, action items or questions — only a one-line summary** — Recordings processed before Recording Insights launched only have the basic summary and an unlabeled transcript. Fix: Open the recording → Summary tab → "Generate insights". CRMown AI re-transcribes with speaker separation and builds the insights (a few minutes for long recordings). If the recording is trimmed, restore the original first, then trim again afterwards.
- **The transcript names the wrong person, or shows "Speaker 2"** — CRMown AI only names a speaker when the recording gives evidence (someone says their name, or a name label is on screen). Similar voices can also be merged or split. Fix: Open Transcript and click the speaker chip to rename — it applies everywhere. Renames are kept until the recording is reprocessed.
- **Ask CRMown AI on a share link says "Too many questions"** — Public share-link questions are capped (10 per hour per viewer, 100 per day per recording) so a forwarded link can't run up your AI usage. Fix: Wait and try again, or ask from inside CRMown (logged-in users have a higher limit).
- **The "Trim" button is missing or greyed out** — Trim is only for videos (not screenshots), only once processing has finished (status Ready), and is greyed out while a trim is already running. Fix: Wait for "CRMown AI is generating chapters…" to finish, or for the "Trimming…" banner to clear, then click Trim. Screenshots use Annotate instead.
- **"Retry processing" says to restore the original first** — Reprocessing reads the full original recording and would overwrite the trimmed transcript and chapters with full-length ones that no longer match the video. Fix: Click Trim → "Restore original", reprocess, then trim again.
- **"Quota exceeded" / can't record anymore on Pro plan** — Pro plan caps at 100 active recordings. Failed and uploading recordings count toward the limit until cleaned up. Fix: Delete older recordings you no longer need (this also cleans up the R2 storage). Or upgrade to Max for unlimited recordings.
- **Upload progress bar stops moving partway through** — Network drop, or the API server timed out a multipart part upload. Fix: Stay on the page — the uploader retries automatically on transient errors. If the bar is stuck for 60+ seconds, close the modal (the in-progress recording is aborted) and re-record. R2 multipart staging is auto-cleaned by the worker so no manual cleanup is needed.
- **Share link returns "private" or 403 in incognito** — Recording visibility is set to Private (default for new recordings is Org-wide). Share links only work for visibility = "Anyone with link". Fix: Open the recording detail page, click "Get share link" (this also flips visibility to share_link), then "Copy link". Test in incognito to confirm.
- **Recording too large to upload / browser ran out of memory** — Multi-hour recordings on memory-constrained devices (older laptops, low-end Chromebooks) can stress the browser. The recorder buffers chunks in memory for local playback before upload. Fix: Keep individual recordings under 1 hour on lower-end hardware. For long-form content, record in segments and link them via the description.

## Error messages
- `Your browser does not support screen capture (getDisplayMedia)` — You're likely on Safari < 13, Internet Explorer, or an in-app browser (e.g. webview inside another app). Switch to Chrome, Edge, or Firefox on a desktop.
- `You've reached the screen-recording limit of 25 on the starter plan. Upgrade to Max for unlimited recordings, or delete older ones.` — Starter caps at 25 active recordings (Free: 10). Delete old recordings to free quota, or upgrade for a higher cap.
- `You've reached your Pro plan limit of 100 screen recordings. Upgrade to Max for unlimited recordings, or delete older ones.` — Pro plan caps at 100. Either delete old recordings (which also frees storage) or upgrade to Max for unlimited.
- `Recording exceeds 3-hour maximum` — Hard cap. The recorder auto-stops at 3:00:00 to keep storage and processing costs predictable. Record long-form content in segments.
- `Chunk upload failed (4xx/5xx)` — Network issue or transient server error during chunked upload. The uploader retries automatically. If the failure persists, close and re-record — the in-progress R2 multipart staging is auto-cleaned.

## Plans
Available on every plan (ungated from the Automation suite July 2026). Active-recording limits by plan: Free 10, Starter 25, Pro 100, Max unlimited, Own license unlimited. Deleting old recordings frees quota.