Building Pi, and what makes self-modifying software so fascinating
April 29, 2026
AI Summary
5 min readPi, the minimalist self-modifiable coding agent that has become the engine behind the popular personal AI assistant OpenClaw, was built by a single developer in Austria who got frustrated with existing AI coding tools. Mario Zechner created Pi after becoming disillusioned with Claude Code, which he found increasingly unstable as its team added features and injected system prompts behind his back. "I don't want my hammer to break at a different spot every day," he says. The result is a tool with only four core tools—read, write, edit, and bash—that achieves extensibility not through a plugin system but by allowing itself to be modified by the user. Pi's popularity stems from this self-modifying nature: it doesn't have MCP support, plan mode, or many other features developers want, but you can ask Pi to build those features into its own code.
Why agents don't feel pain
A central insight from the conversation is that agents lack the human capacity to feel pain from complexity. Armin Ronacher, creator of Flask and an early Pi adopter, explains that when a codebase becomes too complex, human engineers feel the friction—they wake up at 3 AM, they struggle to add features, they curse at the screen. This pain is what drives refactors and rewrites. Agents simply do not experience this. They keep adding to the complexity without ever feeling the consequences.
Continue reading the full summary in the app — free to try.
Read Full Summary →Free • No credit card required
Never miss an episode of The Pragmatic Engineer
Get every new episode summarized in your inbox — free, ~5 minutes to read.
No spam. Unsubscribe anytime.
What you'll learn
- 1 (00:00) **What is Pi and Who Built It** - Introduction to Pi, a minimalist self-modifiable coding agent created by Mario Zechner, now powering OpenClaw.
- 2 (01:37) **Mario's Path from Games to AI** - Mario's origin story: from a working-class childhood in Austria to building compilers and eventually creating Pi.
- 3 (03:44) **Armin's Roots: From Hand-Me-Down PCs to Flask** - Armin's journey from tinkering with old computers to creating one of Python's most famous web frameworks.
- 4 (06:56) **The Austrian Tech Trio: How Mario, Armin, and Peter Met** - The serendipitous connections between three influential Austrian software engineers.
- 5 (08:25) **From Skepticism to Belief: The AI Tipping Point** - Both guests were initially dismissive of early AI coding tools, but a specific moment changed their minds.
- 6 (14:37) **The 30-Team Survey: How AI Agents Are Actually Being Adopted** - Armin's research into how engineering teams are using agents reveals a pattern of vacation-driven adoption and quality decline.
- 7 (19:55) **The Automation Bias Trap: Why Agents Don't Learn Like Humans** - A key distinction: agents don't feel pain or learn from mistakes, leading to a dangerous lack of friction.
+ Full timestamped outline available in the app
Show Notes
Brought to You By:
• Statsig — The unified platform for flags, analytics, experiments, and more.
• Sonar – The makers of SonarQube, the industry standard for automated code review
• WorkOS – Everything you need to make your app enterprise ready.
—
Mario Zechner is the creator of Pi, a minimalist, self-modifying AI coding agent, that is the foundation upon which OpenClaw (created by Peter Steinberger) is built. Meanwhile, Armin Ronacher is the creator of Flask, and a longtime user of Pi. The pair are also friends.
I sat down with Mario and Armin for the latest episode of the Pragmatic Engineer Podcast for an interesting conversation about AI and their reservations about it – even though both are heavily invested in building AI-powered tools.
Mario explains why he built Pi, and gives his take on why it has become so popular. Armin walks us through how he uses AI tools, including building a game with Pi, and why he always puts human judgment firmly at the heart of his approach.
We cover the risks of over-automation, the limits of agentic workflows, and why strong engineers with informed judgment still matter. We also get into the challenges of working with code written by non-engineers, and whether open source can withstand a tidal wave of agent-generated code.
—
Timestamps
(00:00) Intro
(07:30) How Mario, Armin, and Peter Steinberger met(15:15) How 30 dev teams use AI agents: learnings
(21:50) The importance of judgment
(24:26) Challenges when non-engineers write code
(28:30) Downsides of over-automation
(32:18) Pi
(48:09) OpenClaw + Pi
(50:54) “Clankers”
(57:32) Open source and AI
(1:00:22) Complexity as the enemy
(1:02:50) Building an AI-native startup
(1:11:52) “Slow the F down”
(1:16:40) MCPs vs. CLI
(1:25:03) Predictions and staying up to date
—
The Pragmatic Engineer deepdives relevant for this episode:
• The impact of AI on software engineers in 2026: key trends