🚀 Studying CrewAI in Ed Donner’s course on Udemy — and created DungeonMancer for generating D&D monsters!
I never thought that learning multi-agent AI systems would lead me to creating a tool for Dungeons & Dragons, but here we are! 🐉
🤖 Discovering CrewAI Through Learning
How it all started: Ed Donner's "Complete Agentic AI Engineering Course" - https://lnkd.in/duSqq2DB opened my eyes to the possibilities of CrewAI. Instead of one big AI model — a team of specialized agents working together!
🎯 The Project Idea: DungeonMancer
The concept is simple but powerful: - Input: "cyber wolf in neon city" - Output: Full D&D monster stat block with balanced characteristics - Magic: 4 AI agents working as a team
🛠️ The Agent Team
CrewAI helped create a team of specialists: - 🔍 Setting Researcher: Analyzes context and world - ⚖️ Stat Designer: Calculates characteristics and balance - 📖 Narrative Writer: Creates descriptions and lore - 🛡️ Balance Validator: Checks quality and logic
The coolest part? Each agent has its own role, goal, and "background" in YAML configuration. They really work like a development team!
💡 Why This Matters
Before CrewAI, I would have tried to cram everything into one prompt. Now I understand the power of specialization — each agent does what it does best, and the result is much higher quality.
Example result: Neon Stalker — CR 5 cyberwolf with digital camouflage, EMP howl, and the ability to hack neural interfaces. From a simple prompt to a full-fledged opponent in 30 seconds!
🔥 Coming Next
In Part 2, I'll tell you how AI (Grok) helped me write the code for all this — from FastAPI backend to React frontend. Turns out, AI can be a great pair programmer!
Question for the community: Who else is studying agentic AI? What projects are you creating with CrewAI? 🤔
Ed Donner, thanks for the course! Your approach to production AI systems really works 🙏