Hey Nate - just subscribed. Super clear examples, clear walkthrough. It's good to see you make mistakes and leave them in so we can see the real workflow - looking forward to seeing more from you. Great job, bud.
Amazing thank you! :) Could you explain the difference between the system message in the AI agent and the message you write at 12:28 in the vector store tool? Thank you so much
Hi! There’s not too much difference, each tool you want to describe when it should be called, but it’s also helpful to explain this to the agent in its system prompt.
Great job Nate! Re: n8n/pinecode/openAI stack, how/why did you choose this stack? Asides from the obvious reason (no code and fast), how practical is this for the next 6-12 months? Should I prototype with this stack and use langchain and nemo guardrails when the project becomes more serious?
I think n8n and OpenAI will be solid options long term. As for Pinecone, it’s currently the best/cheapest vector store I’ve played around with, but would consider switching as better softwares present themselves. Does that help?
@@nateherk Appreciate your thoughts on Pinecone. For internal staff facing applications (i.e. HR app), agreed that n8n/openAI is a great combo moving forward especially as there are alot of community templates meant to be plug and play, and for prototyping in general. Side note- for something that's meant for public facing, I can't imagine not using Nemo Guardrails to harden the dialog to prevent bot abuse. Nemo Guardrails is avail as an integration through llama index and Nemo Guardrails is designed to work with the langchain framework.
Hi Nate. Great tutorial. Trying to upload not just one file but 100 files to Pinecone. How would such a process look like? It would be such a waste of time downloading each file one by one from Google Drive and then uploading to Pinecone :)
Hey! Two solutions that come to mind without having to code: 1) having all the documents in a folder and reading that folder into an n8n workflow that would upload all of them into Pinecone. 2) creating a looped workflow that would iterate over the download/upload nodes until everything in your drive is in Pinecone. I haven’t actually tested doing this with hundreds of documents so let me test this out and I will look at making a video about this topic!
When you’re configuring the “chat trigger” node, you can choose to embed the widget. Then you will try to chat with the agent and open up the docs where it will give you the JSON script to embed in the website!
Click on the chat button in your workflow to chat with the agent, then at the bottom click on the red "More Info". This will pull up code to embed into a website as well as more documentation on this topic.
G i just wanted to say that your doing a great job at this
Appreciate that!
Hey Nate - just subscribed. Super clear examples, clear walkthrough. It's good to see you make mistakes and leave them in so we can see the real workflow - looking forward to seeing more from you. Great job, bud.
Thank you so much!
Thanks to made this tutorías!. Finally I have found someone who I can follow
Thanks so much!
Awesome... Next should be voice chat RAG... good for practicing and drilling a certain segment of someone trying to learn a new language!
Thanks for your feedback! I have some projects in line, but will definitely add this to my list!
Thank you very much for this Great Content! ❤
Thank you!
Amazing thank you! :) Could you explain the difference between the system message in the AI agent and the message you write at 12:28 in the vector store tool? Thank you so much
Hi! There’s not too much difference, each tool you want to describe when it should be called, but it’s also helpful to explain this to the agent in its system prompt.
Great job Nate! Re: n8n/pinecode/openAI stack, how/why did you choose this stack? Asides from the obvious reason (no code and fast), how practical is this for the next 6-12 months? Should I prototype with this stack and use langchain and nemo guardrails when the project becomes more serious?
I think n8n and OpenAI will be solid options long term. As for Pinecone, it’s currently the best/cheapest vector store I’ve played around with, but would consider switching as better softwares present themselves. Does that help?
@@nateherk Appreciate your thoughts on Pinecone. For internal staff facing applications (i.e. HR app), agreed that n8n/openAI is a great combo moving forward especially as there are alot of community templates meant to be plug and play, and for prototyping in general. Side note- for something that's meant for public facing, I can't imagine not using Nemo Guardrails to harden the dialog to prevent bot abuse. Nemo Guardrails is avail as an integration through llama index and Nemo Guardrails is designed to work with the langchain framework.
@@howardwong2181 interesting. I don’t know too much about that. I will definitely be checking that out. Thanks for the comment!
Hi Nate, can you please make a video on self-hosting n8n on a fully free server like Google for newbies? Thanks.
Will definitely look into this. Thanks for the feedback!
Lots of opportunity in Tutorials for Newbies even if a seperate channel is dedicated for them.
Hi Nate. Great tutorial. Trying to upload not just one file but 100 files to Pinecone. How would such a process look like? It would be such a waste of time downloading each file one by one from Google Drive and then uploading to Pinecone :)
Hey! Two solutions that come to mind without having to code: 1) having all the documents in a folder and reading that folder into an n8n workflow that would upload all of them into Pinecone. 2) creating a looped workflow that would iterate over the download/upload nodes until everything in your drive is in Pinecone. I haven’t actually tested doing this with hundreds of documents so let me test this out and I will look at making a video about this topic!
How can i connect this RAG agent to a test website as chatbot widget?
When you’re configuring the “chat trigger” node, you can choose to embed the widget. Then you will try to chat with the agent and open up the docs where it will give you the JSON script to embed in the website!
@@nateherk for some reason i don’t that option, none in-fact, is it because im on a free trial ?
@@macaronidream that could be it yes, but I’m not 100% sure. I will look into it this week.
Click on the chat button in your workflow to chat with the agent, then at the bottom click on the red "More Info". This will pull up code to embed into a website as well as more documentation on this topic.
What is RAG?
Retrieval Augmented Generation. Ended up changing the name of this video after I filmed it, sorry about any confusion here!