Original video in Chinese.
Key Takeaway
- n8n AI generation: the official “Build with AI” button lets you describe requirements in natural language, and AI automatically searches for, drags in, and connects nodes to form a workflow (such as a daily AI news digest).
- Configuration and flow: credentials still need to be set manually, such as News API/Telegram ID; example flow: 8am Trigger, fetch from 3 sources, summarize top 5/generate images, merge and send to Telegram + chat subflow.
- Conditions and trend: limited to the cloud version/Latest Beta; as models improve and costs fall, the technical barrier drops, allowing people who understand the business to directly customize and implement AI workflows.
AI automatically generating workflows really is the trend. Now even n8n’s official version supports it!
Two months ago, I made an episode of video about AI automatically generating workflows. Back then, it was still pretty troublesome. You had to use MCP to connect Claude to an extra n8n knowledge base, and you also needed a whole set of prompts to remind it and constrain it.
Now the official product has added this feature itself.
When you create a new workflow, you’ll notice a new button called “Build with AI.” Click it and a sidebar opens, where you’ll see a familiar input box. Then all you need to do is tell the AI in natural language what workflow you want to create.
For the demo, I chose the system-recommended “Daily AI news digest.”
After starting it, you can see that n8n’s AI mainly does two things:
First, based on the requirements, it searches for which nodes can be used and drags all of these nodes onto the canvas.
Second, based on the logic and the details of the nodes, it connects all the nodes to form a workflow.
The whole process is pretty fast. After it’s done, the AI will clearly tell you which parameters need to be set manually, and these are generally related to credentials. For example, in this workflow, we need to manually set two things:
One is News API or RSS Feed, which is defining the news source so that AI can fetch it.
The other is the Telegram Chat ID, because the final output is going to Telegram.
Let’s take a look at the workflow process together.
At the starting point, AI sets a trigger for 8:00 a.m. every day. When the time comes, AI fetches news from three sources and then aggregates them.
The aggregated content is then handed over to AI for further processing. The model responsible for generating text summarizes the top 5 news items. The model responsible for generating images creates images based on the news content.
In the end, all of this content is merged and sent to Telegram together.
Since the requirement also includes further discussion of these news items with AI, there’s also a small workflow below for chatting.
Just like I said earlier, although AI can build workflows, we still have to handle some parameters and credential settings ourselves. We can directly double-click the nodes with prompts to configure them, or open them one by one from the list in the dialog box—AI lists them all, which is pretty thoughtful.
This feature in n8n has only been live for a few days, and it hasn’t been rolled out fully yet. To use it, there are two requirements:
First, right now only the cloud version can use it. Whether you’re a free user or a paid user, you can try it.
I think the local version will probably support it later too. I’m actually looking forward to that. Just imagine, in the local version, we could configure our own models—that would be super convenient.
Second, remember to switch the version to Latest Beta in Settings. The default Latest Stable version does not have this feature.
Right now, models’ coding capabilities are still improving, and costs keep going down. For example, the latest Claude Haiku 4.5 achieves 90% of the effect at one-third the price of Sonnet. So going forward, a lot of our highly customized needs that require frequent changes can be handled by workflows built automatically with AI.
Technology is no longer the barrier. Letting people who understand business, understand needs, and have frontline experience directly implement AI is definitely the direction platforms like n8n are heading in.
OK, that’s all for this episode. If you want to learn about AI, want to become a super individual, and want to find like-minded people, come join our newtype community. See you next time!