This is what was to be expected at some point. More or less automatic creation of workflows. Now let’s be quick to utilise this in the turbo speed advancing AI sector. Good luck everybody an make sure you leave me a like and a nice comment. This is my gift to all of you and n8n for the great tool.
Of course some of you may be angry as you have invested a lot of time to learn n8n intricacies. But don’t fret: there is still enough to do with bug fixing, for which you need the expert knowledge. And now we all can build much faster!
Sure, I can share it here, but you might want to consider using X as it involves using Grok. I did give this text prompt to Grok3, the new AI from xAI:
"“Create a workflow for n8n that reads my newest email and if it is about an appointment draft an answer and check my calendar for avialable slots. Use credentials as set up for this workflow” and gave it the JSON of a workflow that I had created earlier that included accessing Gmail and OpenAI. It created a JSON which I saved as a text file and imported to n8n and sure enough the complete workflow manifested. Of course there is still things to fix, but it did create a full workflow with seven nodes that seemed all to make sense. But you can check out the workflow yourself and let me know what you think. Here is the code:
this is already possible before grok 3 came out… there is a custom gpt called n8n superuser (or something to that effect) that can do this… now I am not sure if it does it as well as Grok 3, but either way, its cool and a great place to start no matter which LLM you use
I can’t believe nobody has put together an intuitive and friendly n8n workflow generator so far. I’m not saying it’s super easy but the use case and productivity boost seems obvious to me and the impact on n8n adoption can be significant.
It should be a standard feature at some point, even if initially developed by a third party. Maybe n8n founders are working on it secretely.
Personally, I’ve been using Claude 3.5 Sonnet but its knowledge is outdated and it requires far too much input from me even for simple workflows (and I use LLMs dozens of times per day for various projects).
Such a tool should be trained on the freshest n8n documentation and working examples (JSON workflows).
One of the issues here is that when you’re googling for an “n8n workflow generator” that leverages AI, you’ll mostly find resources related to integrating AI into an n8n workflow because of the way Google processes the query.