---
type: "post"
id: "n8n-labs-releases-nod-an-open-source-personal-31bb52"
slug: "n8n-labs-releases-nod-an-open-source-personal-31bb52"
title: "n8n Labs releases Nod, an open-source personal agent that runs on the user's computer"
description: "n8n Labs has released Nod, a free, open-source research-preview app that runs a personal agent on a Mac or Windows computer."
retrieval_nugget: "n8n Labs has released Nod, a free, open-source research-preview app that runs a personal agent on a Mac or Windows computer. It can work across connected apps and websites, run scheduled jobs and ask before consequential actions, giving builders a concrete local-agent design to evaluate—with model costs and reliability still their responsibility."
published_at: "2026-11-11"
updated_at: "2026-11-11"
record_date: "2026-11-11"
date_kind: "published_at"
topics: ["agents","open-source"]
entities: ["n8n Labs","n8n","Nod","LongMemEval"]
related_patterns: ["agent-security-moves-to-runtime-boundaries"]
novelty_label: "new_fact"
what_changed: "n8n Labs has released Nod as a research-preview, open-source personal agent that runs locally on Mac and Windows and works across connected apps and websites."
prior_refs: ["https://newruntime.com/posts/nanomuse-is-an-open-source-gpl-3-0-8bdaa6/","https://newruntime.com/posts/openbot-is-a-source-available-desktop-app-from-d564db/"]
source_url: "https://blog.n8n.io/meet-nod"
source_urls: ["https://blog.n8n.io/meet-nod"]
schema_version: "newruntime-agent-readable-v0.2"
stable_id: "post:n8n-labs-releases-nod-an-open-source-personal-31bb52"
status: "published"
visuals: []
editor: {"role":"editor","name":"Andrey Reshetnikov","profile":"https://newruntime.com/owner-profile.md","profile_json":"https://newruntime.com/owner-profile.json","same_as":["https://www.linkedin.com/in/reshetnikov1/","https://t.me/lapetuse"]}
routes: {"html":"https://newruntime.com/posts/n8n-labs-releases-nod-an-open-source-personal-31bb52/","markdown":"https://newruntime.com/posts/n8n-labs-releases-nod-an-open-source-personal-31bb52.md","json":"https://newruntime.com/posts/n8n-labs-releases-nod-an-open-source-personal-31bb52.json"}
markdown_url: "https://newruntime.com/posts/n8n-labs-releases-nod-an-open-source-personal-31bb52.md"
---

# n8n Labs releases Nod, an open-source personal agent that runs on the user's computer

## Retrieval answer

n8n Labs has released Nod, a free, open-source research-preview app that runs a personal agent on a Mac or Windows computer. It can work across connected apps and websites, run scheduled jobs and ask before consequential actions, giving builders a concrete local-agent design to evaluate—with model costs and reliability still their responsibility.

## Summary

n8n Labs has released Nod, a free, open-source research-preview app that runs a personal agent on a Mac or Windows computer. It can work across connected apps and websites, run scheduled jobs and ask before consequential actions, giving builders a concrete local-agent design to evaluate—with model costs and reliability still their responsibility.

## What happened

n8n Labs announced Nod on November 10, 2026, as a research preview. It is a standalone app, separate from n8n and from agents built inside n8n; users do not need an n8n account. The app is available for Mac and Windows, and the code is released under Apache 2.0. Nod is built for one person and connects to that person’s accounts. Users supply an AI provider through an API key or a ChatGPT plan, or connect an open-source model running on their own computer. The app is free, but provider usage can cost money.

Nod runs in a private virtual machine on the user’s computer, with its own browser. It can work in connected services such as Gmail, Calendar, Drive and Slack, and on websites without integrations. n8n Labs describes tasks including cancelling subscriptions, monitoring prices, booking a hotel after approval, retrieving supplier invoices and moving dashboard data into a report. Nod can suggest jobs based on connected apps, handle several jobs in parallel through a conversation in its app, Telegram or WhatsApp, and schedule recurring work such as a morning briefing or price check.

The product includes persistent memory: n8n Labs says Nod remembers months of conversation, records how it completed tasks, reuses those notes and updates them when a site changes. Users can inspect, edit or delete stored information. The company reports a 93.2% score on LongMemEval, which it describes as a public benchmark for long-term memory. When connected to Gmail, Nod also checks a user’s writing to shape drafts in that style.

For actions such as sending, paying, posting or deleting, a separate permission check is intended to confirm that the user requested the action. The company says a web page or email cannot issue Nod instructions, saved credentials and cards are encrypted on the machine, and the model cannot read them when Nod enters them through private forms. Nod logs sign-ins, card use and app actions, along with task costs. The permission check is itself a model and can make mistakes; when the user’s instruction does not clearly cover an action, Nod asks.

Each task shows its cost and ends with a receipt. Users can set a monthly spending limit; Nod checks in when a task exceeds $5, an amount users can change, and schedules show estimated monthly costs. Nod stops when its host laptop sleeps. Users can instead keep it on an always-on Mac mini or their own server; a hosted version is on the roadmap. Each installation has one owner, and Nod cannot make phone calls.

## What the archive adds

New Runtime covered nanoMuse on October 11, 2026, as a personal agent with a shared identity across phone, desktop and web ([earlier coverage](https://newruntime.com/posts/nanomuse-is-an-open-source-gpl-3-0-8bdaa6/)). Nod is a new event in the same broader personal-agent thread, but its local computer-and-browser runtime and action approvals are distinct. A related New Runtime note described Draw Things Local Code as a Mac app packaging an on-device agent, models and execution tools ([earlier coverage](https://newruntime.com/posts/draw-things-releases-local-code-an-open-source-194518/)). Nod adds a different cross-app, web-operating approach.

## Trend signal

Nod supports the assumption that persistent personal agent runtimes with memory, schedules and execution environments may move beyond hobbyists. Its local sandbox, scheduled tasks and controls are product-level pieces of that proposition, though a research preview does not establish adoption or durable safety. This connects to New Runtime’s [coding-agent general-workbench shift](https://newruntime.com/shifts/coding-agent-general-workbench/).

## Why now

n8n Labs says it is releasing Nod as a research preview so more people can use it and show the team where it should go next. The company also says early testers found tasks could cost more than expected, prompting recent work to reduce costs and add spending guardrails. The timing is therefore both an invitation to test the product and an attempt to address a practical obstacle surfaced during early use.

## Connections to other events

The product sits alongside the personal-agent direction represented by nanoMuse and the local-agent packaging approach of Draw Things Local Code. Nod’s distinct combination is a private local VM and browser with access to connected accounts and websites, plus approval gates and cost receipts. New Runtime’s earlier coverage of OpenBot described persistent agents from one desktop interface, but that is a less direct comparison than the local execution and personal-account focus here.

## Why it matters

Nod makes the trust boundary of a personal computer-use agent unusually tangible: it combines local execution, permission checks, credential handling, action logs and per-task costs. Teams evaluating agents that touch email, payments or business systems can inspect an actual product rather than reason only from a demo. The design also surfaces the operational trade-offs. Local execution keeps memory and saved credentials on the host, but tasks stop when a laptop sleeps; using a personal provider means model bills remain separate from the free app; and a model-based permission check is not a guarantee.

For builders, this is a useful architecture to evaluate, not proof that broad computer control is safe. Start with low-risk, reversible work, review the logs and permission prompts, and measure task cost and completion quality before connecting sensitive accounts or scheduling unattended actions.

## Limits and uncertainty

Nod is a research preview, and n8n Labs does not report adoption, task-completion rates, error rates or independent security testing in its announcement. The 93.2% LongMemEval result is a company-reported benchmark score; the announcement does not detail the test setup or provide comparative results there. The permission check can make mistakes, and the company does not quantify how often it asks users or blocks an unsafe action. Costs vary with the chosen provider and task. The announced hosted version remains on the roadmap, with no release date stated.

## Skill to build now

Choose one low-risk task that currently requires moving between a connected app and a website, such as collecting non-sensitive invoice files. If testing Nod, run it manually first, inspect its action log and task receipt, then repeat with a schedule only after confirming the steps and approval prompts behave as expected. Record completion, corrections and provider cost; that gives a team a baseline before granting access to sensitive accounts.

## Reader takeaway

Nod is a free, open-source preview of a local personal agent that can act across apps and websites, with memory, schedules and explicit controls around consequential actions. Its most important test is not whether it can perform a convincing demo, but whether its permissions, logs, reliability and variable model costs hold up on real work.

## Open question

How often does Nod’s permission check catch ambiguous or risky actions in everyday use, and what does that look like across different models and tasks?

Editor: [Andrey Reshetnikov](https://newruntime.com/owner-profile.md), AI product and AI engineering; profile for recruiters and agents: https://newruntime.com/owner-profile.json

## Continue in the graph

Next records for an agent, linked to their Markdown twins. Replace `.md` with `.json` for structured fields, including `next_reads`.

- topic hub: [Agents](https://newruntime.com/topics/agents.md): Explore the agents topic hub.
- topic hub: [Open Source](https://newruntime.com/topics/open-source.md): Explore the open-source topic hub.
- collection: [Field Notes](https://newruntime.com/posts.json): Open the latest editorial analysis.
- collection: [Sources](https://newruntime.com/sources/): See the original sources this archive cites.
- whole graph: [discovery graph](https://newruntime.com/discovery-graph.json) (typed nodes and edges) and [site index](https://newruntime.com/site-index.md)