·pro
Book a call
Back to Writing Let's Talk v0.2: Now Running on LangGraph 1.0+ 🚀

Let's Talk v0.2: Now Running on LangGraph 1.0+ 🚀

3 months ago 4
AI & LLMsOpen Source & Releases

Big news: Let's Talk v0.2 is live — and it's running on thedataguy.pro

right now.

This release is a major modernization milestone. The entire stack has been upgraded to the latest stable versions of LangGraph, LangChain, Pandas, FastAPI, and Svelte — making Let's Talk one of the first open-source RAG chat systems to ship on the LangGraph 1.0+ platform.

What Is Let's Talk?

Let's Talk is an open-source AI chat system built on a RAG (Retrieval-Augmented Generation) pipeline. It ingests content from websites and documents, indexes them into a Qdrant vector store, and lets you have grounded, context-aware conversations — powered by GPT-4o-mini / GPT-4.1.

It's what's running under the hood at thedataguy.pro

to answer questions about the site's content.

What's New in v0.2

⚡ LangGraph 1.0+ & LangChain 1.0+

The headline upgrade. LangGraph jumped from 0.6.x → 1.0.x and LangChain from 0.3.x → 1.0.x. These are breaking-change upgrades that bring:

  • A stable, production-ready agentic framework API
  • Improved agent state management and graph execution
  • Better streaming and interrupt support
  • Cleaner developer ergonomics across the board

The frontend similarly moved from @langchain/langgraph 0.4.4 → 1.0.15 and @langchain/langgraph-sdk 0.0.107 → 1.8.0.

📦 Full Stack Modernization

Every major dependency got a version bump:

Package From To
LangGraph 0.6.x 1.0.x
LangChain 0.3.x 1.0.x
Pandas 2.3.x 3.0.x
FastAPI 0.116.x 0.135.x
Svelte 4.x compat Svelte 5
Pydantic 2.11.x 2.12.x
Qdrant client 1.15.x 1.17.x
RAGAS 0.3.x 0.4.x
uvicorn 0.35.x 0.42.x

🐳 Production-Ready Docker Setup

v0.2 ships with a polished Docker Compose configuration for production — including:

  • Health checks on all services
  • Nginx with dynamic configuration templating
  • Custom entrypoint with environment variable loading
  • Pre-built images (no more build-time surprises)

🔧 Code Quality & Reliability

  • Refactored agent variable naming and JSON references (fixes a subtle but impactful bug)
  • Enhanced vector store loading with better error handling
  • Improved code readability throughout the backend
  • Updated GitHub Actions to latest versions (checkout v6, setup-node v6, docker actions v4+)

Try It Live

The upgraded Let's Talk is running right now at 👉 thedataguy.pro

Ask it anything about AI, RAG, D365 F&O, or any content on the site. It's grounded, fast, and running on a freshly upgraded LangGraph 1.0 graph.

Get the Code

The full source is open on GitHub:

👉 github.com/mafzaal/lets-talk

PRs, issues, and stars are all welcome. 🌟

Get in Touch

Have questions about the Let's Talk architecture, want to deploy your own instance, or interested in a RAG-powered chat for your own site or enterprise?

Connect with me:

Whether you're looking to deploy Let's Talk yourself, discuss the LangGraph 1.0 migration, or explore RAG consulting, I'd love to hear from you!