Different halves of the same question
Ask Claude to reason about a decision and it will do it well, from whatever you put in the window. Ask it what your customer said in March and it has no way to know.
Nod holds that half. It captures the call from your Mac's audio, keeps the transcript and summary, and answers across every session with the sources attached.
- Knows your conversations — Nod: yes, the ones it captured · Claude: only what you paste in.
- General reasoning and writing — Nod: limited to your material · Claude: its actual strength.
- Joins or records calls — Nod: captures from your Mac, no bot · Claude: no.
- Together — the connector puts Nod's meetings inside Claude, read-only.
What the connector does
Add Nod as a Claude connector and your meetings join the same chat as everything else you have connected. Claude can then combine what was said on a call with the rest of your context. The grant is read-only and revocable.
Details on Nod for Claude. Nod's own security posture — no stored audio, EU hosting, no model training — is on the security page.
When you don't need Nod
If your work does not run on conversations, Claude on its own is plenty. Nod is for people whose commitments and decisions get made out loud and then vanish.
Frequently asked questions
Can Claude sit in on my call?
No. Claude does not join meetings or capture audio. Nod does the capture, from your Mac's own audio and without a bot, and the connector makes the result readable in Claude.
Is my transcript sent to Claude automatically?
Only when you ask for it in a chat where the connector is installed, and only summaries and transcripts. Nothing is pushed on a schedule and nothing is shared until you install the connector yourself.
Is my data used to train a model?
Not by Nod and not by our subprocessors — our language-model calls run with Zero Data Retention and training disabled. What Anthropic does with a conversation in Claude is governed by their terms, so check those for the Claude side.