Видео

Thanks for great info!!

What are the pros / cons of using PGVector for the vector store?

Super nice Video Cole. I thank you so much for these tips. Now I have a clearlier view at 2025.

With Cursor o1 is built right in. These tools seem pretty similar! I think I'll stick with Cursor since I'm so used to how it works now but Windsurf looks promising.

I regret not having found this video before, how can I get this n8n workflow?

Let me start off with that your channel is great. Very helpful and informative. The question that I have is based on my personal experience with locally self-hosted Supabase. Locally hosted instances of Supabase suck. Not that Supabase sucks. But locally hosted Supabase instances only allow for a single project and no direct access to the settings, including the api, through the Studio UI. If you are using this method for agent document retrieval, wouldn't pgvector be a simpler approach? How are you setting up multiple documents (N8N, Crawl4AI, PydanticAI, etc) in order to allow agents to only access the specific docs and not have them digging through a huge db of various docs that may not be relevant to the query?

Have you at least also considered the possibilities of LightRag or even KAG systems? Thinking of them a lot as well, If there might be a chance to impement them too?

Do you do your improvements after a generation also with bolt.diy or would it better to switch to windsurf, roo-cline, cursor etc. with it?

Can the scraper download the pdfs on a website to add to its knowledge base?

Error with ollama. when I ran your code, I get the following error in ollama object Issues: There was a problem loading the parameter options from server: "Credential with ID "eOwAotC7AUgJlvHM" does not exist for type "ollamaApi"." I saw no where where credentials are needed. Can you please

qwen ai sir in bolt diy

Awesome! Any chance we get this to Agentic RAG in n8n?

That's insane! There any possibility to use it on n8n Workflows?

We just got access to a powerful open source LLM 🐋. Your project will definitely get a steroid boost. Always nice to see people working on projects like this. Wish you luck.

bro do you recommend self hosting n8n?

dark mode homie, my eyes are getting blown out

great video really helpful. I am a new n8n user after following your video's, I was wondering if you could do a video on how to implement an Agentic RAG using N8N.

Not to be funny, but where exactly are the md files saved?

Last time I checked Asana requires a credit card to use for real applications. How much are your master classes videos tied to Asana? I am not interested in paying any subscription or cash to Asana. Do we need to pay anything to use Asana for ALL of your videos? Thank you.

I kinda understand there's a lot of amazing and easy.

Thanks for this.. what would you change if you wanted to use a higher batch size?

najs

Hey Cole, great demo; trying to get this working with Ollama locally but unable to get the model(tried quite a few) to call the .tool function(s). Ever heard of this being done/working? ta c

There was an error processing your request: An error occurred.

cool, thanks how do you handle the "update everyday" rather than a rescrap the whole thing ?

How you're not using Coolify is blowing my mind. All of this could be done as well there as long as you have a domain. This is easier and more straightforward, but coolify is just that awesome ;) Good stuff man, keep up the hard work!

You gotta drop a link to that table/bookshelf combo😂

I've got an idea for a full-stack orchestrator with n8n, but not react and supabase, would you like to talk about it? :) I know that we might need about 15 agents in total :D.

will there be regular compiled together set-up without docker i cant handle docker due to memory over usage am using LPDDR cant upgrade i find hard to have docker based project on while running other projects as well

Now to figure out what to do for those sites that don't have a sitemap

I dont know why you didnt replied to every comment expect mine 😢...this was the comment (Bro in Bolt.Diy when i upload the same file that i downloaded from bolt.new or any other source code file , it is getting imported in bolt.diy but when i try engange with it showing something went wrong error. iam using deepseek llm v3 and even tried openAI's llm too but same issue any idea to solve it ?)

i want to see zapier and make out of business. n8n for the win. open source for the win. control your data or join the matrix.

Thanks for the easy to follow tutorial! I changed the .env encryption and secret keys during install and get this error when running n8n "There was a problem loading init data: Credentials could not be decrypted. The likely reason is that a different "encryptionKey" was used to encrypt the data." Do I need to update the keys in another location?

This is exactly what I was looking for - except I was hoping to add either text or voice prompts from users to ask about Google Sheet information. I wish you included how to code in a voice prompt and the necessary settings :(

Awesome video! Thanks for knowledge sharing!

What makes this an AI Agent? It seems Agents tools/frameworks are everywhere*, but not sure what capabilities does using Pydantic AI have over prompting existing AI tools (deepseek, chatgpt, perplexity, etc)? *Phidata/Agno+Streamlit, Browser-Use , AgentQL SmolAgents , PydanticAI , LlamaIndex , LangGraph , CrewAI , AutoGen, EidolonAi , Swarm etc Any thoughts on Brave Web vs Tavily (Pathway / LlamaIndex)? I'd like to point an AI to a YT video (ex. JRE 2237) , that DOESN'T have transcription, and have it answer questions about it. Or upload a datasheet, and have it write a driver for it. Is something like this only possible with Agents?

this is amazing, what about, using the data that the LLM has aquired, to then use it to code on cursor, or so, how would we go about it? Thank you for ur amazing work!

will it work with supabase?

could u do this with sharepoint where there are several pdf, words, excels? So u could ask it where can i find the weight-balance files of this specific helicopter for example Or an excel wich contains all the KPI's and u could ask it

can I use qdrant self hosted instead of supabase?

finnaly i found you

Thanks!

This is great. Thank you for sharing.

Hey Cole, great tutorial and I learned a lot. Just wondering if I want to isolate documents by user (they only can see their own doc), what should I do in the vector table? Something like user id?

That's excellent but...🤔 Is it necessary to create a local network first? Or is the network already created with the installation? or is it no longer necessary...?

I wonder how to scrap all the links if the website does not have a sitemap... Is there any way of doing it with Crawl4ai?

I would like to know how to use a local file instead of google drive, like maybe point it to an obsidian folder, or it would be nice if it can point to a folder on another computer or nas system

this vid should be called "how to use Docker to deploy programs" .. but your headline is more clickbaity

Great content, really appreciate what you do. Can you please make a video showing how to run this workflow with crawl4ai locally installed in Docker. Thank you Cole. Keep going!