By Vlad Kampov, Nod co-founder · Published September 1, 2026
- Product evidence
- Nod produces transcripts, structured summaries, decisions, action items, and cross-session answers with source sessions. Capture is bot-free and audio is not stored.
- Limits
- Nod is macOS only. Team context is early access, and Nod does not execute work in email, documents, CRMs, or project tools.
- Method
- This page is based on reproducible workflows in the current Nod macOS product and the public security, pricing, and product documentation linked below. It makes no aggregate usage or outcome claim.
Sources and related reading
The five parts of working memory
Working memory starts with conversation memory, turns each call into meeting memory, and makes the result available through Ask Nod. Optional ambient context adds the text around the conversation. Team context can make selected material available to an approved early-access team.
- Capture the conversation without adding a meeting bot
- Extract decisions, commitments, action items, and open questions
- Ask across several sessions and open the source behind an answer
- Keep personal material private unless you deliberately share it
A reproducible workflow
Capture a customer call, a product review, and a voice note in Nod. Put the related sessions in one folder. Before the next call, ask: What did we decide, what did I promise, and what is still unresolved? Open the cited sessions, check anything consequential against the transcript, then transfer the confirmed work into the system where it will be executed.
What working memory is not
It is not perfect recall, an employee, or an automatic system of record. Nod can miss context and generated answers can be wrong. It does not send email, edit documents, update a CRM, or post to Slack. Its role today is to capture, organize, retrieve, and show the conversational evidence so a person can decide what happens next.
Explore the pillars
Start with conversation memory for continuity across calls, meeting memory for the durable record of one meeting, AI coworker for the human-in-control boundary, and AI context layer for using meeting context inside other assistants. For the difference from a recap, read meeting summaries vs working memory.
Frequently asked questions
What is AI working memory?
AI working memory is a searchable, source-linked record of decisions, commitments, and context that remains useful across conversations rather than ending with one summary.
How is working memory different from meeting notes?
Meeting notes describe one call. Working memory connects many captured calls and lets you ask a new question across them, with the source sessions available for checking.
Does Nod act on my behalf?
No. Nod captures, organizes, remembers, and answers. It does not send email, edit documents, update a CRM, or post to team tools.