AI chatAsk the meeting, not your memory
Ask anything about the meeting — what was decided, who owns which action item, what to put in the follow-up email. The full transcript, your notes, and persistent memory are all in context.
Everything in context
Every chat thread has the full transcript, meeting notes, knowledge attachments, and participant details available. Answers reflect what was actually said — not what the AI infers.
Summaries and follow-ups on demand
Turn a 60-minute call into a structured summary, a list of action items, and draft emails — in one prompt. No rewatching, no reconstructing from notes.
Memory that compounds
After each meeting, Scriba distills key context into persistent memory. Start the next chat already knowing who owns what and what was decided — without repeating yourself.
AI chat — common questions
What AI model does Scriba use?
Managed mode uses Anthropic Claude for chat and summaries. No API key is required — Scriba handles the connection. A BYOK option for OpenAI GPT-4o is available in Settings > Advanced for users who need it.
Is my meeting data used to train AI models?
No. Transcripts are processed through Scriba's managed pipeline and are not used to train any model. All meeting data stays in your local archive on your Mac.
How many AI messages do I get?
Free includes 50 managed AI messages per 30-day window. Pro includes 1,000/month. Max is higher. You can add credit packs on Pro and Max if you need more before the period resets.
Can I chat in a different language?
Yes. Scriba has a per-meeting output language setting — you can get summaries and chat responses in a different language from the transcript.
What is meeting memory?
After each AI reply, Scriba distills the conversation into a persistent memory for that meeting. When you open a new chat thread for the same meeting, key context — decisions, names, action items — is already there.