Jan 16
Loading video player...
Everyone’s talking about *agentic AI* right now... but very few people can actually show you what that means in real products. In this tutorial, we’re going to build a real multi-agent system from scratch: an AI Tweet Reply Bot that routes mentions to specialized agents based on sentiment and context. By the end, you won’t just know the buzzwords. You’ll know how to architect your own agent systems. You can find all prompts here: https://getlightswitch.notion.site/Agentic-Tweet-Bot-Prompts-2b0788c13245801f97acecb172c4fae1 *What we’re building* In this video, we’ll build a full-stack AI Tweet Reply Bot that: - Listens to incoming tweets or mentions - Uses a **Router Agent** to classify sentiment (compliment, support, troll) - Routes to specialist agents that respond with the right tone - Runs on a **FastAPI** backend and a **React** frontend - Uses **LangGraph** to orchestrate the entire multi-agent workflow *Chapters* 00:00 – What is “agentic AI” and why it actually matters 05:18 – Step 1: Full-stack setup in Cursor (FastAPI + React) 09:17 – Step 2: Installing LangGraph and building the first agent 14:15 – Step 3: Designing the router + specialist agents on paper 19:30 – Step 4: Building the multi-agent system in LangGraph 23:07 – Step 5: Testing real tweets (compliment, support, troll) 24:22 – Wrap up + what to build nex If this helped you understand agentic AI beyond the buzzwords, hit subscribe so you don’t miss the next build. I post weekly tutorials on building real AI products, especially for indie builders and small teams trying to go from “vibe-coded prototype” to “actual product that makes money.” 👉 Drop a comment with the next system you want to see: support inbox, Slack bot, content router, something else? #AgenticAI #LangGraph #AItutorial #FastAPI #ReactJS #IndieHackers #AIDevelopment #VibeCoding #OpenAI #SunnySaysYes