What it does This repository contains 52 Jupyter notebooks demonstrating how to build generative AI agents at increasing levels of complexity. Each tutorial pairs runnable code with a free companion explainer on the DiamantAI site. The progression runs from simple conversational agents with memory, through framework introductions (LangGraph, MCP), up to domain-specific systems for customer support, academic planning, scientific literature review, and data analysis.

The interesting bit The author treats this as an interconnected curriculum rather than a grab bag of scripts. There are explicit “related projects” for memory techniques, RAG, prompt engineering, and a separate production-readiness repo — suggesting someone actually thought about where a learner should go next. The recently added entries (HR AI Assistant, Art Tourguide with LightRAG, Contextual Quoting System) show the collection is still expanding.

Key highlights

• 52 tutorials with step-by-step documentation and runnable implementations
• Covers LangChain, LangGraph, PydanticAI, and MCP frameworks
• Free companion explainers at diamant-ai.com/tutorials (89 total, per the site)
• Explicit difficulty progression: beginner → framework → educational → business use cases
• Community contribution workflow via CONTRIBUTING.md
• Related repos for memory, RAG, prompt engineering, and production deployment patterns

Caveats

• The README is heavily interleaved with book promotions, job board banners, and newsletter signup CTAs; the actual tutorial list requires scrolling past substantial marketing material
• “52 tutorials and growing” appears in one place, but the visible table truncates well before that number; you have to open individual notebooks to see the full scope
• No visible dependency management file (requirements.txt, pyproject.toml) in the README excerpt; setup friction is unclear

Verdict Worth bookmarking if you’re past the “hello world” LLM stage but still figuring out how agents compose in practice. Skip it if you need production-hardened reference architectures — the author explicitly routes that audience to the separate “Agents Towards Production” repository instead.