Chatting with 44K Fashion Products: LangChain Opportunities and Pitfalls

In this video, I’m going to uncover some LangChain pitfalls and opportunities by building a fashion e-commerce chatbot from scratch using generic building blocks with data from Hugging Face.
Link to the code:
colab.research.google.com/dri...
▬▬▬▬▬▬ V I D E O C H A P T E R S & T I M E S T A M P S ▬▬▬▬▬▬
0:00 Introduction and overview
0:48 AI and Headless Ecommerce
2:20 Chatting with 44K fashion products
5:15 What do we want to achieve?
7:15 Pitfalls and opportunities

[@farzinhaddadpour7192]

I loved this video! Short and very informative! Thanks for the effort!

[@rabbitmetrics]

Thank you! Glad it was helpful

[@aimademerich]

Phenomenal thank you 🙏🏽

[@avidlearner8117]

Fantastic video. Anyone working on an actual product can relate to this. Yes, there's a ton of "how to" content, with live code that give incredible results. But as soon as you go beyond the demonstration, the real work starts and you actually change the way you see LLMs. They suddenly become mega functions and a lot less magical. Still! Plenty of stuff to discover. But one jey aspect I think is separating everything into agentic tools and having a lot of small schemas to map every little aspects so that the larger LLM can then have a complete picture of a lot of small elements. Thanks for sharing!

[@BunniesAI]

Agreed. Awesome video 🤩 I’m running an e-commerce store with 90.000 products (car parts and gear for cars). Finding the right part for the right car would be tremendously helpful in a conversational style (we have a car-database, but it’s not the interaction I think the customer wants). Getting past the “simple” stuff and into the nitty gritty would be awesome.

[@rabbitmetrics]

@NyeMedier2010 It sounds like a case where you might need to customize the BaseRetriever class used in the conversational chain. Appreciate the comment - thanks for watching!

[@rabbitmetrics]

Much appreciated! Thanks for watching. Will dive into the creation of good tools at some point.

[@BunniesAI]

@@rabbitmetrics I’m signing up for your community 🤩

[@rabbitmetrics]

@@BunniesAI We would be delighted to have you join

[@haojiehang5863]

Hey there, wondering what tool is used for build the presentation, the animation and overall flow looks pretty good.

[@rabbitmetrics]

Thanks! I'm using Final Cut Pro for the "explainer part" of the videos.

[@kumarj693]

I am not a Coder but I work as a Linux Engineer. Honestly, I didn't really understood the concepts that you explained completely, can I still join your rabbitmetrics course for learning AI & its use cases & be able to create my AI business where I can make extra bucks ? I am interested in learning AI & implement it but I have no idea where to start.

[@rabbitmetrics]

@kumarj693 Thanks for watching. The goal with the community is to bring together the tech and business side of AI and focus on solving problems. We're only starting up now, but non-coders have joined already.

[@altered.thought]

doesn't open ai function calling solve the structured response issue? as per this tutorial -> ruclips.net/video/8eCZeFhvyGE/видео.html

[@rabbitmetrics]

Yes, that would one way to solve it but it ties you to OpenAI. I will revisit this issue in later videos. Thanks for watching

[@Poprollable]

Causal ai is the solution. In the meantime, knowledge graphs.