Adding RAG to LangGraph Agents

[@glowstorm334]

Appreciate your videos very much , Please do more videos on Langraph , your videos are my guide to learning Langraph.

[@samwitteveenai]

Thanks glad they are useful.

[@kenchang3456]

Congrats Sam! This is the most detailed video that I have experienced from you and I really appreciate the time it must have taken to develop and edit the project. Yes please extend the project as I'm sure I will continue to learn more from your sharing. Best regards to you.

[@samwitteveenai]

Thanks Ken

[@GriffinBrown-tq9jz]

Yoooooooooooo!!!!! This is the best ai channel. Feels like I’m getting mentored by an expert

[@anhhct]

You should consider writing a book since your way of explaining things is concise and easy to understand :)

[@TheGenerationGapPodcast]

The best way to get Reliability from Agentic workflow is use an orchestrater like Prefect for the part of the process that must be deterministic. Don't justleave it up to the LLM decision engine to call 3 functions for a particular workflow that consists of 4 tasks that must be a part of a dependency graph. Just call a flow that will call the functions in the right order. Human in the middle can also be implemented in the Orchestrator. We then need to use conditional edges for flow control, kinda "like if-else statements".

[@theh1ve]

Awesome content as always and great point around logging Q/A pairs to find gaps in the RAG knowledges!

[@dixon1e]

Thank you for taking the time and effort to do this and help us all.

[@andy111007]

A perfect example of Recursive Reprompting: (a) prompting a general-purpose language model to construct a structured overarching plan, and (b) generating story passages by repeatedly injecting contextual information from both the plan and current story state into a language model prompt. We then revise by (c) reranking different continuations for plot coherence and premise relevance, and finally (d) editing the best continuation for factual consistency....... I hope you can add reranking to it in future examples

[@kai_s1985]

Thank you very much! This is awesome and it is a great example for using AI in our everyday lives! Now, we can automate our e-mails to our customers, and it is indistinguishable from personalized e-mails by humans.

[@darwingli1772]

This is really useful! Thanks! Another popular use case is RAG Chatbot. It would be great to see how langgraph can be implemented to achieve sth like the OpenAI assistant (maintain conversation thread/routing questions) but with more flexibility!

[@CK.23.]

Great job. Very interesting. Thanx. And yes, of course, pls add the Research Router..

[@rhiteshkumarsingh4401]

wow very well explained. this and the previous video u mentioned in this one.

[@brandonheaton6197]

The new features in Autogen, particularly the local vector database for Teachable agents (the new subclass for ConversableAgent) is awesome and deserves a video Maybe build out a teacher assistamt that fetches youtube transcripts, writes NGSS based quizes from the transcript, and posts them to google drive

[@samwitteveenai]

Agree the teachable agents is cool I will get around to making some vids.

[@attaurrehman3233]

Hey Sam, appreciate your videos. I've learned a lot from this langchain series. Could you make a video on langGraph agents using the tools and some custom tools to complete the task?

[@NicholasNicolo]

Also curious about this. Is there a reason you aren't having the agents use the tools?

[@pj5776]

Could you please make a end to end project ( uploading pdf or doc and getting information from that) using open source LLM, using streamlet as a frontend part. Please make a detailed video on that. And Thanks for making such videos, those are really helpful.❤

[@codekiln]

Can you discuss the utility of the sections like `|start_header_id|>system`?

[@realCleanK]

Thank you!

[@seadude]

Is LangChain/Graph ready for PROD yet? When I was last researching, the syntax/library had breaking changes often.

[@samwitteveenai]

Its a lot more stable now. There are still some issues and it depends on your project and the model you are using whether I personally would say it is a good fit.

[@stanTrX]

Dear Sam i am new and trying some experiments. For the agents, rags and tools, where shall i start for today? Langchain or langflow? Which will be more convenient? I may also integrate memory etc later on but my initial priority is tools such as search etc and rag. Thanks.

[@samwitteveenai]

Get to know LangChain and basic RAG stuff first then go for LangGraph later

[@TooyAshy-100]

Thank you

[@PRColacino]

The link to Colab is wrong ;)

[@samwitteveenai]

there were 2 now just the correct one.

[@PRColacino]

@@samwitteveenai tnks Sam !

[@sperazza]

your content is great, very helpful and useful. Good job.