Pi Agent Creator on future of Agentic Coding

In this episode, I sat down with Mario Zechner, the creator of Pi Agent - one of the most talked about coding agents in the developer community right now. Mario isn’t your typical AI founder. From compiler engineering to graphics, game engines, and applied ML and now deep into LLMs and agents, his journey brings a unique lens to where programming is headed. If you're a developer trying to understand where things are going then this conversation will genuinely expand how you think about coding. We go deep into: The story behind building Pi Agent What “agentic coding” actually means (beyond the hype) His experience working with AI-assisted coding over the last 2 years Why current dev workflows are fundamentally broken The shift from writing code → orchestrating intelligence The future of programming and developer roles European startup ecosystem vs global ecosystem His open-source journey and philosophy The long-term mission behind Pi Agent Timestamps: 00:00:00 – Intro 00:00:45 – Coding all nighters and Agentic addiction 00:05:30 – His experience with AI coding over the last 2 years 00:08:44 – OpenClaw moment and his perspective 00:12:45 – Why agentic coding sucks 00:17:20 – Agentic coding without human in the loop 00:19:52 – Agent harness vs model capabilities 00:35:20 – Future of programming 00:48:00 – Amp is the real testing lab 00:53:00 – Why OpenClaw used Pi Agent at core layer 01:01:30 – Future of Agentic Coding 01:02:30 – Amp Code Shoutout 01:06:28 – Mario on working with Frontier Labs 01:21:00 – What keeps Mario awake now 01:26:28 – WTF moment for Mario in AI so far Available on Spotify : https://open.spotify.com/episode/0XXziOlvDRgYMkiFRGabGl?si=UqTcySY_SLq6PWsw9EC0BA Follow Mario on X: https://x.com/badlogicgames Follow me on X for more conversations like this: https://x.com/waghnakh_21