How to LangChain + Flowise + Pinecone + Render = Chat With EVERYTHING!

Рет қаралды 21,658

Жыл бұрын

Join us in this deep dive as we unravel the power of LangChain LLM agents, Flowise - the cutting-edge visual LLM tool, Pinecone - the game-changer vector store DB, and the world-renowned OpenAI.
Harness the future of artificial intelligence and transform your digital world! We'll navigate through LangChain LLM agents, revealing how they are revolutionizing the AI landscape. Then, we'll unlock the potential of Flowise, a visual LLM tool that's making waves in the tech ecosystem.
Github: github.com/FlowiseAI/Flowise
Render: render.com/
But, we're not stopping there! Get ready to explore Pinecone, a dynamic vector store DB that's redefining the way we handle data. Finally, we'll demystify OpenAI, the pioneer that's shaping the future of AI technology.
Whether you're an AI enthusiast, a tech-savvy coder, or a curious mind, this video is your key to understanding the interconnected world of advanced AI tools.
Remember, the future is not something that happens to us; it's something we create. So, let's create it together.
Don't forget to hit that like button, share with your friends, and subscribe for more insightful content. Get ready to embark on an exciting journey through the world of AI! 🚀🔥
#AI #LangChain #Flowise #Pinecone #OpenAI #FutureTech #LLMAgents #VectorStoreDB

Пікірлер: 35

@DerekJoinerPRESTONEZ Жыл бұрын

Just made it to 6:06 and wanted to say I appreciate the in depth walkthrough. I needed the "explain like I'm stupid" approach.

@ChrisCollett-Plus Жыл бұрын

Hey man, this is a great tutorial. Just what I was looking for. Cheers!

@DreamsAPI Жыл бұрын

Thank you sir for a very beginner friendly with the end objective of having a line app that will be useful. Appreciate your videos, can't wait for more

@TestSavant Жыл бұрын

Thank you very much for the feedback!

@jd2161 Жыл бұрын

What's a line app?

@turbolike7 Жыл бұрын

5 Star Tutorial! Thank you!

@bendingreality- 11 ай бұрын

Awesome, thanks for sharing. Would it be possible to not only upload PDFs individual but make a broader dataset 'searchable' by Chatgpt? ie. pulling data from a CRM, Trello, voice recordings, Zoom recordings etc.

@tomspaccialbelli6630 Жыл бұрын

Hello, I'm seeking clarification on how to display the outcome of separate prompts within a single chat session. For instance, if I first prompt a summary about a company and then subsequently ask for three additional paragraphs, how can I ensure that the results for all these prompts - the summary and the three paragraphs - are shown collectively within the same chat, even though they were initiated separately? Any guidance you could provide would be greatly appreciated. Thank you.

@youwang9156 11 ай бұрын

appreciate ur video, thank you so much !!!! just i got one question. it doesn't matter about our input length or whatever, the dimension of the pinecone always stays the same for 1536?

@swannschilling474 Жыл бұрын

I was just installing Flowise Yesterday and was hoping to connect it to a local LLM, any idea on how to do this?

@wsligter Жыл бұрын

Great tutorial, thanks. One thing I'm not clear on is when/how/how often to populate the vector database. Having 'upsert document' in a conversation flow would mean it's updated on every turn - too slow. Having it populate once and then moving to the 'chat with vectors' flow for the conversations means it's populated once and then never again - not ideal for web scrapes as websites change regularly. Any ideas / pointers on this? what's a good way to set this up?

@lancerkind 5 ай бұрын

You’re entering the realm of the data pipeline. A data pipeline runs a series of operations and you decide how often frequently you need to run it daily hourly weekly monthly. I don’t know if flowise has that capability. If it doesn’t have that capability, then you need to use another tool to execute the flow that Flowise creates. Behind the scenes, Flowise creates Javascript code. You can find that code, you can execute that code with your data pipeline tool, which will run, as frequently as you program it to. If you are familiar with build pipelines, data pipelines are the data version of that. I have to expect that flow wise. I have to expect that Flowise has this capability since everyone who builds chat bots will eventually need to update their data.

@Francotujk 4 ай бұрын

Hello! If my app needs to get the pdf from the user (the user will upload the pdf in a web app). How can I define metadatas (to only query the llm with the pdf uploaded by that user? Thanks!

@heirey Жыл бұрын

Thank you for these incredible tutorials! I have a question: every time I ask a question, the whole flow is reused from scratch, so I find it quite expensive. Isn't it possible for pinecone to keep the data so that "text-embedding-ada" isn't used every time?

@wiktormigaszewski8684 11 ай бұрын

maybe there are some caching solutions?.. just guessing

@OnyxStudiosInteractive Жыл бұрын

That was a great tutorial! Thanks for creating it. I had a question... have you tried any of the other vector stores like Chroma or Weaviate? Also they really need to find a way to save flows so apps and embeds work in production. Thanks a lot!

@DreamsAPI Жыл бұрын

Hey Thomas, you can use your own servers to host flowise or langflow, then depending upon on your preference you can expose your bot, I plan on using the curl option

@OnyxStudiosInteractive Жыл бұрын

@@DreamsAPI Thanks! I'm using Railway right now for Flowise and it's working pretty well. It's just that the devs are still working on the Chroma integration.

@sandowssecret7363 Жыл бұрын

Thank you for this awesome video there is so little information out there about flowwise. It's great how you explain every little aspect of it. I know zero coding so the deeper your explanations the better. I was wondering if their is a way to save the message history somehow?

@wsligter Жыл бұрын

@@DreamsAPIcan you elaborate how you expose your bot using curl? Is it a module in flowise or does it require custom code?

@yusukejohnson5083 11 ай бұрын

im unable to type my metadata i.e., "doc_type : "pdf"" "id : 1" etc. and have it saved. Any way you guys can help?

@Eagleiris Жыл бұрын

will you make that template available to download?

@liuchuchenliu2802 Жыл бұрын

I have been using the azure openai api to do langchain on flowise for the past two days, but it has not been successful. I would like to ask if someone has done it

@urmatallatra Жыл бұрын

Thank you for the content you provide, you do a very cool job, I do not know English, I watch your videos through a translator, I would really like to learn from you how you can upload data in other languages. Maybe there is a way before uploading to the server to use API to translate the text from the document from Russian to English, then when all processing is finished, output to the chat in Russian. Perhaps there are even better solutions to enter in another language and output the information in the same language.

@tim0d 8 ай бұрын

Can you please point me to a tutorial where we can input an image and use Hugging Face or Replicate to convert the input image to a text description of the image?

@mudassarm30 10 ай бұрын

Doing it with OpenAI API is very easy, but it is so expensive. Can you do it using only huggingface embedding and LLM models ?

@lancerkind 5 ай бұрын

He solves the problem with pinecone later in the video.

@Nuninecko Жыл бұрын

Thanks for the interesting video. As a SUGESTION, would you find interestin explore how to have Flowise agents do stuff on our behalf given a goal/app detailed pages on Flutterflow?

@TestSavant Жыл бұрын

Thank you for the feedback and the suggestion! I have a video on the subject that im planning now that you might find interesting

@AI-in-Action952 9 ай бұрын

how can we limit the Gpt model from answering unrelated to questions, i want to limit it only to my data. in flowise

@vanessajambois 9 ай бұрын

Give it instructions in the System Message, found in Additional Parameters of the Conversational Retrieval QA chain. For example: You will provide me with answers from the given documents and URLs. If the answer is not found, say exactly "Hmm, I am not sure how to answer that. Try asking another question." and stop after that. Refuse to answer any question not about the info. Never break character.