This course teaches you to build production-grade AI agents through ten complete, full-stack projects. Rather than stopping at simple chatbot demos, you will ship real applications with authentication, databases, background jobs, vector search and streaming user interfaces, all written in TypeScript. The course is organized around the most important agent frameworks in use today. With the OpenAI Agents SDK, you will build a multi-agent customer support platform with tool calling, human escalation, and input and output guardrails. You will then build an AI travel planner that syncs itineraries to Google Calendar. With the Claude Agent SDK, you will create a Cursor-style coding assistant for the terminal and an autonomous GitHub pull request reviewer that indexes entire codebases.
Using LangChain and LangGraph, you will build a RAG-powered chat for GitHub repositories and a NotebookLM-style study notebook that answers questions with citations from PDFs, documents and websites. You will also build a visual workflow automation platform inspired by n8n. With the Vercel AI SDK, you will develop a ChatGPT clone and a Perplexity-style answer engine featuring live web search, citations and image generation.
In the final module, you will write your own agent SDK from scratch. You will implement tools, memory, planning, execution, workflows, streaming, tracing, guardrails and multi-agent handoffs, and learn how agent frameworks work internally. Along the way, you will gain hands-on experience with Next.js, Prisma, Drizzle, Neon Postgres, MongoDB, Better Auth, Clerk, Inngest, Pinecone and React Flow. This course is designed for developers who are comfortable with JavaScript or TypeScript and want to move beyond prompt engineering to build reliable, real-world agentic applications.
