# n8n Labs releases Nod, an open-source personal agent that runs in a local sandbox

Canonical: https://theagenticweekly.com/news/n8n-labs-nod-open-source-personal-agent · Evidence: Reported · First seen: 2026-11-10 14:15 UTC · Source published: 2026-11-10 12:52 UTC · Brief published: 2026-11-10 14:58 UTC · Updated: 2026-11-10 14:58 UTC · Author: Agentic Weekly

> n8n Labs released Nod on Nov 10, 2026 as a research preview: a standalone, open-source personal agent that "runs on your computer" with its own sandbox and browser. n8n says it scored 93.2% on the LongMemEval long-term memory benchmark.

## What changed

- n8n says Nod is a research preview from n8n Labs, a standalone app for Mac and Windows that is separate from n8n and needs no n8n account, server or cloud service to set up.
- Nod works inside a private virtual machine on the user's computer with its own browser, acting in connected apps such as Gmail, Calendar, Drive and Slack and on websites with no integration.
- The code is open source under Apache 2.0, and users connect their own AI provider with an API key or a ChatGPT plan, or an open-source model running locally; the app itself is free.
- n8n says a separate model-based check confirms the user asked before Nod sends, pays, posts or deletes anything, saved logins and cards are encrypted locally and entered through forms the model cannot read, and every action is logged with its cost.
- Cost guardrails include a per-task receipt, an optional monthly spending limit, a check-in once a task passes $5 (adjustable), and an estimated monthly cost for each scheduled job.
- Known limits per n8n: Nod stops when the laptop sleeps, each installation has one owner, it cannot make phone calls, and a hosted version is only on the roadmap.

## In the source's words

> "Nod is a standalone app, separate from n8n, that runs on your computer."
> (n8n blog, Meet Nod, https://blog.n8n.io/meet-nod/)

> "The permission check is itself a model and can make mistakes, which is why Nod asks when your words don't cover an action."
> (n8n blog, Meet Nod, https://blog.n8n.io/meet-nod/)

## What we have not verified

- We did not install or run Nod; everything here is n8n's description.
- The 93.2% LongMemEval score is n8n's claim; the post does not link a full evaluation setup and we did not reproduce it.
- We did not audit the sandbox, the permission check or the claim that a web page or email cannot give Nod orders.
- n8n says the latest release cut task costs substantially but gives no figures; real costs depend on the model provider chosen.

## Timeline

- Source published: 2026-11-10 12:52 UTC
- First seen by our watcher: 2026-11-10 14:15 UTC (via Watcher: n8n blog feed)
- Brief published: 2026-11-10 14:58 UTC

## Sources

- [n8n blog: Meet Nod](https://blog.n8n.io/meet-nod/) (primary source) (fetched 2026-11-10)
- [Nod on GitHub (n8n-io/nod)](https://github.com/n8n-io/nod) (primary source) (fetched 2026-11-10)

## Related

- [n8n: verdict and setup guide](https://theagenticweekly.com/tools/n8n)
- [Alternatives to n8n](https://theagenticweekly.com/alternatives/n8n)
- [All news briefs](https://theagenticweekly.com/news)
- [New AI models this week](https://theagenticweekly.com/models/this-week)

## How to cite

Agentic Weekly, "n8n Labs releases Nod, an open-source personal agent that runs in a local sandbox", published 2026-11-10 14:58 UTC, https://theagenticweekly.com/news/n8n-labs-nod-open-source-personal-agent