Help & Support

Click New Recording in the sidebar. Your browser will ask for microphone permission \u2014 grant it. Then click Start Recording and speak. To capture audio from a Zoom or Teams call, share your screen with audio enabled when the browser prompt asks how to capture.

Any modern browser that supports the Web Audio API: Chrome, Edge, Firefox, Safari (macOS). Mobile Safari and Chrome on Android also work but the recording UI is optimized for desktop. No installation, no extension, no Zoom plugin.

When you click Start Recording, choose the screen-share option from the browser prompt. Make sure to check "Share audio" at the bottom of the browser's screen-share dialog \u2014 this is the key step. Scribe will capture the system audio output instead of your microphone.

Yes \u2014 click Upload Recording in the sidebar. Scribe accepts MP3, WAV, M4A, WEBM, and most common audio formats. AWS Transcribe will process it and you'll get the same summary, structured notes, and visualization features as a live recording.

Every 45 seconds while you're recording, Scribe sends your running transcript to Anthropic's Claude model (via AWS Bedrock) and asks it to rebuild structured notes from the full conversation. That's why you see Recap, Decisions, Actions, and Key Points populate in full sentences \u2014 it's not just keyword extraction, it's the model genuinely understanding the conversation.

AWS Transcribe's speaker diarization identifies different voices but doesn't know their real names. After the recording ends and processing completes (typically 3-5 minutes), you can rename each speaker by clicking the label in the transcript. We're working on auto-detecting names from context.

Yes. Click any text in the transcript and start typing. You can correct misheard words, fix punctuation, or remove filler content. After editing, you can regenerate the AI summary so it reflects your changes.

Three on-demand AI artifacts:

• Entity Graph \u2014 force-directed visualization of every person, company, and concept mentioned, with relationships between them.
• Architecture Diagram \u2014 if your meeting describes a technical system, this generates a Mermaid flowchart of it.
• Jira Tickets \u2014 your action items reformatted into proper Jira tickets with description, acceptance criteria, priority, and labels. Click Copy on any ticket, paste into Jira.

The summary is generated by Claude based on the full transcript. Accuracy depends mostly on transcript quality \u2014 if AWS Transcribe missed a word or attributed it to the wrong speaker, that can cascade. Always read the summary before relying on it for decisions. Especially for high-stakes content (medical, legal, financial), verify against the original transcript.

On the Pro plan, yes. Open any meeting and click Enable Sharing. You'll get a public read-only link you can paste anywhere. The shared page shows the summary, transcript, and visualizations \u2014 not your account or other meetings.

The Print button at the top of any meeting opens a clean printable view. Use your browser's "Print to PDF" option to save it. We're working on direct markdown and DOCX export.

• 3 meetings per month
• 45 minutes per meeting (120 minutes total per month)
• Live transcription with speaker labels
• AI summary & structured live notes
• 3 each of: Entity Graphs, Architecture Diagrams, Jira Ticket sets, Ask Your Meetings queries
• 7-day meeting storage

See the pricing page for the full comparison.

You'll see an inline message ("3 of 3 Entity Graphs used this month"). Click Upgrade to Pro to unlock unlimited use. Limits reset on the 1st of each month \u2014 you don't lose your existing meetings, just can't generate new ones until the reset or until you upgrade.

Click Upgrade to Pro anywhere in the app, or open Settings > Plan & Billing. You'll be redirected to Stripe to enter payment details. Your account is upgraded immediately on successful payment. 30-day money-back guarantee on your first Pro subscription.

Settings > Plan & Billing > Manage Subscription opens the Stripe customer portal where you can cancel. Cancellation takes effect at the end of your current billing cycle \u2014 you keep Pro access until then. Or just email hi@usescribe.io.

Check three things:

• Browser permission: click the lock icon next to the URL and make sure microphone is allowed
• System input device: your OS sound settings might be using the wrong microphone
• Mute button: ensure your physical mic isn't muted on a headset or webcam

The audio level indicator in the recorder shows real-time input \u2014 if it's flat, the mic isn't reaching the browser.

AWS Transcribe usually finishes within 3-5 minutes for a 30-minute recording. Longer recordings can take up to 15 minutes. If your speaker labels still haven't appeared after 30 minutes, the Transcribe job may have failed silently \u2014 email us with the meeting date and we'll investigate.

Free plan caps recordings at 45 minutes. You'll see a warning at 40 min and 44 min before auto-stop. Upgrade to Pro for unlimited-length recordings. If you're on Pro and a recording stopped under 45 min, that's a bug \u2014 please email us with the date and time.

Audio playback uses presigned S3 URLs that expire after 60 minutes. If you've had the meeting open longer than that, hard-refresh the page (Ctrl+Shift+R) to get a new playback URL. Meetings older than your plan's retention window will have their audio deleted permanently.

Two options:

• Open the meeting, click Regenerate Summary from the Summary tab \u2014 sometimes a second pass picks up things the first missed
• Use the floating Ask Claude chat to ask a specific question about the meeting \u2014 it'll search the transcript and answer with context

Settings > Danger Zone > Delete My Account. You'll need to type your email to confirm. This permanently deletes all your meetings, transcripts, audio files, and account data within 24 hours. There is no undo.

All data is stored encrypted on AWS in the United States. Audio files in S3 (encryption at rest), transcripts and metadata in DynamoDB. AI processing happens via Amazon Bedrock with Anthropic's Claude model, which does not retain prompts or train on them. See the full Privacy Policy for details.

Recording laws vary by US state and by country. Some states (California, Florida, Illinois, etc.) require all-party consent. You are responsible for obtaining consent from everyone in the conversation before recording. Scribe shows a consent reminder before each recording but cannot enforce the law on your behalf. When in doubt, just announce: "I'm recording this for note-taking" at the start of the meeting.