🚀 New Tutorial: Build a Document Chatbot with Streamlit, ChromaDB & Mistral (Available in English & Spanish!)

Show the Community!
1

Hi Streamlit Community! :waving_hand:

I’m excited to share a fantastic video tutorial that walks you step-by-step through building a robust Retrieval-Augmented Generation (RAG) document chatbot. Whether you are just starting with AI apps or looking to level up your Streamlit projects, this is a highly practical resource.

What You’ll Learn in this Tutorial:
The video showcases how to seamlessly stitch together a modern AI stack using a Streamlit frontend. Key highlights include:

  • Streamlit UI: Building an intuitive interface for uploading documents and a chat interface for querying them.
  • ChromaDB Server Mode: Moving beyond local persistent clients to set up and connect to a standalone Chroma database server.
  • Mistral AI: Utilizing Mistral’s API for both generating embeddings and powering the conversational LLM.
  • Smart Document Parsing: Using Microsoft’s powerful MarkItDown library to clean and convert various file types (PDFs, DOCX, PPTX, etc.) directly into Markdown for highly optimized LLM indexing.

:globe_showing_europe_africa: Breaking Language Barriers
One of the coolest things about this resource is its accessibility. The tutorial was originally recorded in Spanish, making it a great native resource for the LATAM and Spanish-speaking developer community.

However, if you don’t speak Spanish, you are completely covered! Thanks to YouTube’s new AI dubbing feature, you can watch the entire video with a high-quality English audio track. Just click the gear icon (:gear:) on the video player, navigate to “Audio Track,” and select English. It’s an amazing way to share global knowledge without language barriers.

:laptop: Get the Code
If you prefer to learn by doing, the entire source code is open-source and ready to be cloned. You can dive straight into the implementation details, see how the Streamlit session state is managed, and spin it up on your own machine.

  • :television: Watch the Video:

Let’s Discuss!
I’d love to hear your thoughts on this stack. Have you been using ChromaDB in Server mode with your Streamlit apps? Have you tried MarkItDown for your RAG pipelines yet?

Drop your thoughts, questions, or share your own implementations in the thread below. Happy coding! :balloon:

2

Thanks for sharing @germancastano . If you have any feedback based on building this in terms of how Streamlit might focus future feature development in this area, please share!

3

Bom trabalho

4

Really useful tutorial. I like the MarkItDown step for handling different document formats, and using ChromaDB in server mode makes the setup more practical for larger projects.