Context -
I am very new to n8n and workflow automation with non-technical background. I am creating a workflow to automate contract management (POC) using free tools available.
Tools in use -
n8n - self hosted
ngrok
google suite
openAI
Process - (test phase)
Drive will contain 4 folders, archive, template, final document, approved document.
The AI agent must process chat input and retrieve the contents from the template folder, modify it using content from archive folder and create a final preview of the document. After confirmation of the modified content create a file in final document folder.
Once the document is successfully created trigger email notification for approval. The approval process will be manual and the approved file will be moved to approved document manually. Integrate with e-sign and trigger mail to the concerned parties.
Execution so far -
I have added the following nodes - chat trigger, ai agent (with descriptive prompt), chat model, window buffer memory, 3 drive node tools and 1 gmail tool.
Issues -
The operation I selected was to ‘create from text’ on drive node. Although the ai agent is able to retrieve the right content from template and modify it, it is not able to create a document in the final document folder with the modified content. Also, it creates empty files in each folder every time tool is called by the agent. I tried using http request but the technical requirement of configuration was getting too complicated so I switched to drive node.
Questions -
- Which operation is suitable to extract data from files in drive for an AI agent?
- Does the current process requirement align with the functionalities of the selected nodes?
- What are some changes in the workflow that can resolve and execute this