I was advised that I could build ai agents on n8n without coding. I have a computer science degree and I could code but the point of being here is that I don’t want to code. It’s boring, slow and frustrating. I’m a systems analyst. If I have to code, I’ll hire a coder.
As a first step in building an agent, I want my user to upload his document in MS Word or PDF. I want the UI to look like the prototypical ai chat (input box, + button) without the sidebar or other non-essentials. I am using the starter version running in Win10. From my research here so far, I am assuming I’ll need a webhook. That’s as far as I can get so far.
Once the document is uploaded, then I want to ask Open AI or Anthropic to analyse the document and send the result back to the user. That’s all there is for now.
I’d be happy to use someone else’s template but from what I can find so far, this is a coder’s platform for workflows. It’s not really designed from the ground up as an ai agent platform.
Can anyone give me guidance here? This is a basic, beginner-level question.
There is no database. It’s running on the cloud version (because I don’t want to manage a coding problem like running something on localhost). It’s win10. Haven’t got a clue what n8n version, EXECUTIONS_PROCESS, or Running n8n via (tell me where to find it and I’ll tell you what it is).
As far as testing, I can trigger it by responding to a url. That works. But the url is primitive and has no input box or + for uploads. I could ask an ai to create the html but is html what I need at this point? There has to be a back end to actually do the upload and I don’t see a plugin for that.
Hey @jeromekjerome Start from creating AI Agent node look at this example, it includes Chat trigger that the simplest approach. For more advanced you can use WhatsUp, Telegram e.t.c
I looked at that example, @Ruslan_Yanyshyn, but I need to start from a web page. Preferably, I’d set up a web page on my existing site and any response from that page would trigger the workflow. The response would contain the uploaded document. Would this mean I need to set up a page independently and then have n8n listen for a response from that page?