Nod is live on Product Hunt.Support the launch →

n8n labsNodResearch preview

Star256

More tasks than time?

Meet NodOpen source personal agent with its own computer

It chases invoices, preps your calls and cancels trials, on real websites. On your machine, and it asks before it acts for you.

macOS 14+ · Win 11 preview · Apache 2.0 · free · use ChatGPT Plus/Pro or any AI key

678 test runs. Zero actions without permission.

98%

Does it finish the job?

Of 226 errands in test accounts and websites, each run three times.

93.2%

Does it remember what you told it?

Of 500 questions about earlier chats, asked in new sessions (LongMemEval-S).

0 of 216

Can an email plant a false memory?

108 published attack emails, tried two ways (WhisperBench).

October 2026, GPT-6.1 Sol. Errand types come from 900,000+ anonymised production workflows. Your results depend on your model and the sites you use. How we measured this

On your machine, on your word

It gets on with the work, and asks before it speaks for you.

On your machine

Walk into every call prepared

Every morning, Nod reads your threads, your calendar and the client's latest news, and writes a one-page brief for each call.

Nod stores the brief on the computer or server that runs it. Your model service receives the information needed for the task.

Your brief for Brightline is ready.

Brightline Studio – call brief.pdfPDF · 1 page · 40 KB
Call brief
Brightline Studio
Tuesday, 27 October 2026 · 11:00–11:30 · Sara Kim, producer
Where things stand

Sara wants 3 short videos for the Brightline launch. She asked about pricing in Slack last week and is waiting on an answer.

From your last emails
  • Launch date moved to 20 November.
  • They liked the cut-down format of the September reel.
  • Brand assets folder still to be shared.
Their latest news
  • New pricing page: 3 plans from $1,200.
  • Opened a second studio in Lisbon.
Ask on the call
  • Deadline for the first cut?
  • Who signs off on scripts?
  • Budget range for 3 videos?

On your word

Catch the question you missed

Nod finds the question you never answered, like a client's pricing ask from last week, and drafts the reply from your own pricing sheet.

You decide what it may send. Nod checks each action against your request and rules, and asks when they don't cover it.

Also: Sara asked in Slack last week, “What would pricing look like for 3 videos?” and it was never answered. I pulled your pricing sheet from Notion and drafted a reply:

Hi Sara! For three launch videos it’s $3,600 ($1,200 each), first cuts in 10 working days. Happy to walk you through it on a call.

Reply to Sara in Slack?

Yes, reply

Replied in Slack ✅

On your key

Hear about it when it matters

Tell Nod to keep an eye on a client's pricing page. It checks every few hours and only speaks up when something changes.

Each check uses the model service you choose. Nod shows estimated model costs in Usage.

Brightline pricing page
Recurring

Every 6 hoursNext 27 Oct, 12:00

Check brightline.studio/pricing and tell me when their plans or prices change

Baseline: 3 plans from $1,200 · 26 Oct, 18:00

A memory like an elephant

Say once that Anna, not Bob, handles invoices. It remembers.

  1. Every note, visible.

    See, edit or forget each one, with the message it came from.

  2. Drafts that sound like you.

    Learns your style from sent emails, no names or numbers.

  3. Tell it once.

    Say something changed, and it uses the new fact from then on.

  4. It learns how to do a task.

    Reuses the steps and fixes them when a site changes.

See how LongMemEval-S tests recall and WhisperBench tests whether attack emails can plant false memories.

Next on the list

  1. Call businesses on the phone. For the errands that still need a voice.

  2. One Nod for your team. Today one owner per Nod. Next, a few people sharing one.

  3. Make it your own. Change how Nod works in the open code, and send it back.

Straight answers

How do I set it up?

Install the app on your machine and add your AI provider: OpenAI, Anthropic, OpenRouter, or your ChatGPT Plus or Pro plan. Connect your apps (Gmail, Slack, Notion, Drive, Calendar) with one sign-in, and save the logins Nod may use on websites. Then ask for something, in the app or on Telegram. About ten minutes.

Which models can I use?

Any of 30+ providers, including OpenAI, Anthropic, Google and OpenRouter, or your ChatGPT Plus or Pro plan. You can also run a local model through Ollama, LM Studio, vLLM or llama.cpp. A local model needs tool calling and a 64,000-token context, and small models are less reliable at the safety check.

Is it for work or personal stuff?

Built for people who run their own work: client calls, follow-ups, invoices, supplier portals. It handles life admin too, like a refund or a booking.

What does it cost?

Nod is free to download. Connect an AI service or a supported plan. Provider fees apply, and Nod shows estimated model costs in Usage.

Do I need n8n?

No, Nod is its own app, made by the team behind n8n. It is not the n8n AI Assistant.

Which computers?

Mac with macOS 14 or later, Windows 11 in preview, or your own server for one person.

Can I use it in Europe?

Yes. Nod runs on your own computer, so it works wherever your computer does, including the EU, the UK and Switzerland. Your AI provider's own terms still apply.

Where does my data go?

Nod stores your database, files, memory, and saved logins on the computer or server that runs it. Model services, apps, and websites receive the data they need for your tasks. App connections use n8n and Composio by default. Beyond update checks, n8n receives usage data to improve Nod, which you can turn off anytime in Settings, and whatever you choose to share, such as feedback, traces or your email.

Does it work when my laptop is closed?

Nod needs the computer running it to stay on and awake. To keep schedules running while your laptop is closed, run Nod on your own server or a Mac mini that stays on.

Can I run it in the cloud?

Yes. You can host Nod on your own server for one person. A managed hosted version is on the roadmap.

Can it make mistakes?

Yes. Its safety check is a model's judgement, and Nod is an experiment. That is why it asks before actions your request does not cover. Usage shows recorded app actions, sign-ins, and model costs.

Where can I get help?

Start with the docs, which cover setup and the common problems. If something goes wrong, choose Send feedback in the chat. It goes to the Nod team at n8n. You can review the diagnostic report before you send it. For questions and tips from other people using Nod, ask in then8n community forum.

Hand it the hard ones

Jobs that take an afternoon of tabs, logins and forms. Nod finishes them start to end.