Build an AI Agent to Remember All Your Notes

Do you wish ChatGPT could remember and leverage all your study materials? This Python tutorial shows you how to create a powerful RAG (Retrieval-Augmented Generation) AI agent that does exactly that!
You'll learn to create vector embeddings from your Markdown notes, allowing the AI to retrieve relevant information with reliable text chunking.
Here's what you'll learn:
👉 Extract content from Markdown notes into logically chunked embedding chunks
👉 Generate vector embeddings using the Google Generative AI API
👉 Implement vector similarity search for retrieving relevant passages
👉 Construct embedded prompts to the Gemini 1.5 Pro LLM
👉 Create a production-ready RAG agent in Python

#ai #python #chatgpt

[UjjwalSidhu]

This is amazing, Really love it when they teach what is what and how it is happening, Thanks for this :D hope to see more from you.

[julienduchesneau]

oh wow Thank you, it did work and that stuff is gold, I have started for fun coding using chatGPT and I can tell you some of my afternoon was just turning in round as chatGPT dont learn and repeat the same mistake. I will try to make some agents from now. Also maybe make an interface instead of the terminal. Great job, can't wait for more

[RoronoaZoro-fqcw]

stoked on more content! i'd love more my man.

[Gatrehs]

Something like this would be sick to implement with Obsidian honestly.

[slackerpope]

I'm a student. GPT hallucinations make that technology nearly useless. I need to know I can trust what I get back from a prompt. This method seems like a good answer. Please make more videos to help students and other folks who need facts and reliability in their AI tools. Example, I want to be able to ask my documents questions and be absolutely sure the answers I get back are ONLY coming from that document.

[jarad]

Awesome I don't want to have to build it though just use it is or available?

[mattiharo]

Nice! Could you provide corresponding example using Spring AI and Java, utilizing RAG pattern?

[hashemtarawneh]

LM studio has the Local model and the embedding LLM and it works as an openai API and you can load more than one model at the same time, can you make the same project with LM studio

[viper]

RAG is great for “needle in a haystack” type questions, but struggles with “many needles and mini haystack type questions” making it not great for certain scenarios. Would be interested to see a workflow to fine tune on the RAG data so it could later be eliminated and a new VectorDB stood up to cache new data until the next round of fine tuning.

[rabeemohammed]

Can you make a video about creating a chat program with files that will be local without the Internet?

Thank you for your wonderful effort and excellent explanation in conveying the information.

[dot]

why didnt you use Llama3? it‘s said to be superior..

[agu]

Vamos a ver que invento nos traes hoy 😊

[newssir]

Bro, but you did not test it to show that it really remembers everything.

[uldisbriedis]

Bro, nice stuff, but too much AI generated text in your video. It's hard to folow, humans dont write (speak) so long sentences, add some prompt for human syntax writing style for text to make it more natural 😉

[johnjakson]

So I can build a charming female AI agent, this Associate in India won't be a scammer after my money will it?

[charliepersonalaccount]

No idea what the hell he's talking about lol. Too technical for me

[uhtexercises]

Another RAG tutorial. Why not just call it that? Agent has become such a buzzword in the AI space.