Loading video player...
In this video, I share a hobby project I built with LangGraph to automate repetitive YouTube post-production work. The tool is an AI agent that helps with tasks like proofreading captions, generating chapters, writing descriptions, suggesting SEO-friendly titles, and creating thumbnail ideas. Instead of using a generic off-the-shelf tool, I wanted something that matched my own workflow and could save time on work that has to be repeated for every upload. I walk through a live demo of the CLI-based app, show how it processes video and playlist metadata, and explain the human-in-the-loop review step before anything is published. I also cover how the workflow is built using LangGraph, LangChain, OpenAI APIs, and YouTube APIs. Later in the video, I talk about how I built the project using AI-assisted development with structured specs, implementation plans, issue tracking, pull requests, and unit tests. I also share the biggest lessons from the project: why specification matters more than coding, why tests are essential for agentic development, and why human judgment is still critical even when AI writes most of the code. If you are interested in AI agents, LangGraph, LangChain, or practical AI-assisted software development, this video should give you a useful real-world example. #LangGraph #LangChain #AIAgents #AgenticAI #AIEngineering #SoftwareEngineering #AIAssistedDevelopment #OpenAI #YouTubeAutomation #CLI #DeveloperTools #Productivity #LLM #Automation #Python #AgenticWorkflow #HumanInTheLoop #SEO #YouTubeCreator #AIWorkflow 00:00:00 - Why I Built a LangGraph YouTube Agent 00:02:55 - Demo: Automating Video Metadata in a REPL 00:06:46 - Demo: Playlist Description and Title Generation 00:08:28 - Human-in-the-Loop Agent Design with LangGraph 00:11:00 - How I Built It with Codex, Copilot, and PRDs 00:15:09 - Lessons Learned: Specs, Tests, and AI Workflow 00:19:06 - Why AI Won’t Replace Software Engineers Follow me on X: https://x.com/anilvdeshpande Official WebSite: https://codetutorhub.dev/