Original video in Chinese.
Key Takeaway
- Workflow for ordinary people: no need to mess with complicated AI workflows like n8n (which has a high barrier to entry for B-end users). Gemini’s new Gem feature (conversational workflow generation) is enough to meet everyday C-end needs. It works out of the box, and you can describe what you want verbally while it automatically breaks things down and builds nodes.
- Gem demo example: enter a keyword (like Robotaxi), and Gem automatically generates a process (research the keyword + Deep Research searches 3-5 English reports/papers + information distillation + a deep analysis report), outputting structured content such as Core Facts/Key Insights.
- Advantages and expansion: fast iteration through conversation (for example, switching to Chinese), backend node editing (changing the model), and seamless native integration with the Google ecosystem. The backend capabilities will definitely keep improving in the future. Simple tasks can all be handled by Gem; for complex cross-ecosystem pipelines (like Notion/Slack/MySQL), use n8n.
For ordinary people, there’s really no need to build complicated AI workflows like n8n. Those are tightly tied to business operations, are suitable for B-end users, and the barrier to entry is pretty high. For individual C-end users, Gemini is enough. That’s because Google recently upgraded the Gem feature in Gemini. Now it can generate workflows. Let me show you.
Here’s my requirement, and it’s very simple. I want Gemini to search the entire web based on the keyword I enter, find the latest 3 to 5 core English reports or papers, then distill the information, weed out the nonsense, and extract the core facts and viewpoints. Finally, it should generate a deep analysis report.
This is the interface for the new GEM, and it has already been switched to a conversational format. After I enter that long requirement, Gemini analyzes and breaks it down, then automatically generates a workflow. Every step is based on my needs, such as first researching the keyword, then doing Deep Research, and so on.
Let’s do a test. I’ll give it a keyword: Robotaxi.
This is a topic I’ve been paying close attention to lately, because I hold Tesla stock. Their Robotaxi business has made major progress on the regulatory side and has started to truly become driverless. That will have a very big impact on the company’s fundamentals and its stock price.
After a short wait, a report comes out. It includes Core Facts, Key Insights, and more.
At this point, you might say: isn’t this basically the same as search?
From a content perspective, if you use Deep Research, you can also get something similar in a chat format. But that content is all in plain text.
On the new GEM side, though, what it gives you is a page, an app. Beyond simple presentation, it can also have other functions, like downloading.
This is what Google wants to achieve, and it’s called a “Mini APP.” In familiar terms, that’s basically Google’s version of a mini program.
Through the workflow, this mini app packages the result into an independent entity. What it delivers is no longer just information, but an entity, or in other words, software.
And this software is editable and changeable. For example, in the demo just now, the output was in English. I can use conversation to make it output Chinese, while the information sources remain in English. A change like that can be done in one or two minutes.
If you think modifying it through conversation isn’t enough and want finer control, you can go into the backend page. At that point, you’ll see the full workflow and can modify individual nodes. For example, you can change the model for a certain node.
With the new GEM and Google’s version of a mini program, most everyday needs for ordinary people can be met.
It has an extremely low barrier to entry, works out of the box, and you can just tell it what you want. And the integration with the Google ecosystem is native, so the experience is the smoothest.
And later on, Google will definitely upgrade this mini app’s capabilities. It won’t just be about front-end presentation; it will also be able to offer more powerful backend capabilities. That is definitely going to happen.
Of course, the drawback of the new GEM, or rather the thing it still can’t do, is cross-ecosystem integration. For example, if you want to build an AI workflow that first stores data in Notion, then sends it to Slack, and finally writes it into MySQL, Gemini still can’t handle that kind of complex cross-ecosystem chain. That’s where n8n is strong.
So simple tasks can all be implemented inside GEM. When you run into a complicated business workflow, then you can consider n8n.
OK, that’s it for this episode. If you want to understand AI, want to become a super individual, and want to find like-minded people, come join our newtype community. See you next time!