Hi everyone,
I’m new to the forum but have been using n8n for a while.
To be brief, I’d like to know how you leverage AI, especially considering one of the biggest problems I’ve faced with most models (ChatGPT, Gemini, Deepseek, Claude): asking them to build a workflow in JSON format is practically impossible. They always get it wrong. This is expected behavior, of course, since most have knowledge cut-offs around July 2024.
In my case, I have to explicitly tell the AI to read the documentation from a specific link. This helps it give me configuration ideas and explain what each option does. For instance, I asked it for help creating Redis nodes and it suggested operation values that don’t exist in the current versions. I decided to give it this link: https://github.com/n8n-io/n8n/blob/master/packages/nodes-base/nodes/Redis/Redis.node.ts. This worked perfectly, and it knew how to help me.
Unfortunately, I can’t always find the up-to-date documentation for some nodes, like the “Merge” node v3.2.
So, to sum it up: what techniques do you apply to get AI to help you with n8n using correct, up-to-date documentation? Just telling it to base its response on version 1.???.? and find the info on its own isn’t enough; it always provides incorrect information unless I give it the direct source.
I would greatly appreciate any tips to improve this situation.
Thanks!
- n8n version: 1.101.2
- Database: SQLite
- Running n8n via: Localhost
- Operating system: Ubuntu 24.10