Building Agentic AI Workflows with Matthew Henage - JSJ 678

JavaScript Jabber

In this episode, we sat down with full-stack developer and AI innovator Matthew Henage, creator of WAOS.ai (Web App Operating System) and the incredible storytelling platform SpeakMagic.ai. This conversation took us deep into the world of agentic AI, low-code app building, and the future of intelligent workflows.
We kicked things off with Matthew sharing how he’s been riding the AI wave since GPT-3.5 blew his mind. His platform WoWs is all about making it easy for developers to build powerful web apps with embedded AI workflows — think of it like Zapier meets ChatGPT, but with agents working together instead of API chains.
One of the most eye-opening parts of our chat was learning about agent swarms — essentially teams of specialized AI agents that collaborate to perform complex tasks. Instead of relying on one giant AI brain to do everything, you create smaller, purpose-built AIs that handle specific steps in a workflow. It’s scalable, smarter, and kind of like assembling your dream dev team… but all made of code.
Matthew’s Speak Magic project is a jaw-dropper. It uses a swarm of over 40 agents to turn a single story idea into a fully animated, two-minute video — complete with scenes, scripts, character animations, music, and more. It’s AI storytelling on steroids.
We also talked a lot about:

• Best practices for building reliable AI workflows
• The importance of keeping context windows small (under 4,000 tokens works best!)
• How prompt engineering is becoming the new programming
• Using AI for vibe coding (yes, that’s a thing) and rapid prototyping
• The tradeoffs between using traditional programming vs. letting AI handle logic
• Ethical considerations and how to handle memory and privacy in long-running user interactions

Check out Matthew’s work at WAOS.ai and speakmagic.ai — and as always, stay curious and keep building!
Become a supporter of this podcast: https://www.spreaker.com/podcast/javascript-jabber--6102064/support.