n8n's developers just shipped their free, open source AI agent that does exactly what Grok Bot, Dots and Muse do. For me, the interesting part isn't the agent, it's the five thigs around it: governance, safety, sovereignty, privacy and customization. Basically where it lives and who sets the rules, specially interesting for all those working with clients who care a lot about regulations, governance or privacy, GDPR...definitely look at this one if that's the case. They have created a separate platform for this with their own branding. It's called Nod, released just a few days ago as a research preview. It has its own computer and its own browser, so it works in your apps and also on websites. You talk to it in one conversation (completely mirroring Grokbot, Dots and Muse), and you use it from the desktop app, a web browser, Telegram or WhatsApp. It is a separate app from n8n (you don't need an n8n account). Why I find it interesting? With this much work in trust and security behind it, I think it's a serious bet on autonomous agents. A bit on context: n8n's developers are from Germany, where GDPR and regulations around AI are HUGE, and in my opinion this approach is great for all businesses, niches and corporations in need of full sovereignty, governance, privacy and customization. Here's where I see the interesting part for everyone outside Europe too: if you can mirror European GDPR regulations, then it will surely be 100% corporate compliant. (Also is specially interesting for European-based people since Dots, Muse and Duet are not available here yet). What I find very interesting here is not the agent itself. It's the five things around it: governance, safety, sovereignty, privacy and customization. - Governance: The model only makes the plan. A separate check on the Nod server decides what actually runs, and it sees only your words and the action, never the email or web page that suggested it. A Usage page shows every action, its cost, and your words that allowed it, and there's a hard monthly spending limit. - Safety: Anything Nod reads from outside is marked as data, (never as instructions) so an email can't approve an action for example. Memory only learns from your own words. Logins and cards sit in a vault the model never sees, and every payment needs your card code. - Sovereignty & Privacy: You pick where it runs (the desktop app, from source, or your own server) and which model it uses: an API key, your ChatGPT plan, OpenRouter, or a local model. - Customization: It's open source (Apache 2.0), so you can read the code behind each decision. It connects to apps, MCP servers, APIs and your files, runs on schedules, and can use parallel helpers. - n8n direct connection: It can also build workflows in your n8n.