I’m wondering if it’s possible to hook up a couple of different types of nodes for a knowledge source for the AI. The first is google drive. In the setup I see that I can set the resource to a folder but the operation is where I don’t see an option to read or index. So, is it possible to set a folder up to be used to reason from?
Also, I see AI nodes, but not an AI tool. A coworker created a GPT in our OpenAI teams account and shared it with me. That GPT knows how to answer customer requests. Is it possible to set up an AI tool that uses that GPT?
Yes, You Can Use a Google Drive Folder as an AI Knowledge Source in n8n — Here’s How:
While n8n doesn’t have a one-click “index folder for AI” feature, you **can absolutely configure it to read documents from a Google Drive folder and feed the content into an AI model like OpenAI for reasoning.
I’m sorry but your how to doesn’t give me enough info to know how to accomplish this. Can you please provide some more details on how to set up the google drive tool to read from a folder and provide that to the AI? Thank you!
To get all the files, use the search operation. You can leave the query blank. It will then return every file. If you want to specify a specifc folder where to get the files you can also specify what folder.
It will return 1 file, as 1 item. So the split in batches node is not needed. The next node in question would loop for each item.
It connects directly to the AI agent. The AI node has an option called “Automatically Passthrough Binary” that sends the data to the agent. It’s enabled it in the example below.
If you would like to give the AI agent all the data in one run. You can add the aggregate node after the 2nd node.
If this helped, it’s appreciated if you mark it as solved. Cheers!
Thank you. I’m sorry but I feel so stupid. I’m just not completely following how this will work as the tool for my ai agent in my current workflow. Here’s what it currently looks like after I added what you have. How do I hook it up so that my existing OpenAI node that’s generating the answer based on the google doc and sheet, also uses the drive files as a tool? And yes, I would think passing all of the files in at once would probably be best, so do I have the aggregate node in the right spot? Thanks again.