Pi-Mono

The AI Assistant Toolkit That Went Viral

The Simple Version

Meet the Creator

🇦🇹
Mario Zechner from Austria. He previously built tools that helped millions of people make video games. Now he's building tools that help AI assistants do coding work.
"Keep it simple."

That motto drives everything about how Pi-Mono is designed — small pieces that work together, nothing more than what's needed.

What is Pi?

Think of Pi like a universal remote control for AI

📺
You know how a TV remote lets you control your TV? Pi lets you control AI assistants. Instead of talking to each AI separately (ChatGPT, Claude, Gemini), Pi talks to ALL of them using the same buttons.
💻
It can read files, write code, run programs — like a super-powered assistant that works with any AI brain you choose.

Why Did It Go Viral?

One project sparked a chain reaction

🚀
Another developer named Peter built something called OpenClaw — think of it as an "AI butler" that lives on your computer. OpenClaw uses Pi as its brain.
145K+
GitHub fans in 1 week
MIT
Tech Review 2026
1 wk
To go viral

That's like a new app getting millions of downloads overnight. It was on the news, podcasts, everywhere.

The LEGO Analogy

Pi-Mono is like a box of LEGO bricks

There are 7 different sets inside, and each one builds something different — but they all snap together.

🌐
pi-ai
The Translator
🧠
pi-agent-core
The Brain
🔨
pi-coding-agent
The Worker
🎨
pi-tui
The Pretty Face
🌍
pi-web-ui
The Web Version
💬
pi-mom
The Slack Bot
☁️
pi-pods
The Cloud Manager

Brick 1: The Translator

pi-ai

🌐
Imagine you need to talk to 20 different people who each speak a different language. Instead of learning 20 languages, you hire ONE translator who speaks all of them. That's pi-ai.

Brick 2: The Brain

pi-agent-core

👨‍🍳
This is like the manager at a restaurant:
1
Takes your order (what you want done)
2
Gives instructions to the kitchen (the AI)
3
Waits for the food (the response)
4
If the kitchen needs something, the manager goes and gets it (tool calls)
5
Brings the finished dish back to you

Brick 3: The Worker

pi-coding-agent

This is the actual product people use — a super-smart coding assistant in your terminal.

You type what you want. It reads your code, writes new code, and runs tests.

Only 4 Basic Skills

  • Read files
  • Write files
  • Edit files
  • Run commands

But with just these 4 skills, it can do almost anything!

Brick 4: The Pretty Face

pi-tui

🎨
Every app needs a user interface. Pi-tui is like a makeup artist for terminal apps. It makes the boring black-and-white terminal look colorful and pleasant.

Bricks 5, 6, & 7

The Web Version, The Slack Bot, and The Cloud Manager

pi-web-ui — The Web Version

A chat interface you can put in any website. Like embedding a mini ChatGPT in your own app.

pi-mom — The Slack Bot

A bot that lives in Slack. Mention @mom and ask it to do things. It remembers conversations, installs software, and schedules tasks.

Fun fact: "mom" has made 18 code contributions to the project all by itself!

pi-pods — The Cloud Manager

For running AI on powerful computers in the cloud. Like a valet parking service, but for AI models.

"Here's my AI model, park it on a powerful computer and keep it running."

How They All Connect

The LEGO bricks snap together like this

pi-ai (translator)
   ↓
pi-agent-core (brain)
   ↓           ↓          ↓
pi-coding-agent  pi-mom     pi-pods
(worker)          (slack bot)  (cloud)

pi-tui (terminal looks)  ↑ makes it pretty
pi-web-ui (web looks)    ↑ makes it pretty

Each piece does one job well. Swap any piece without breaking the others.

The Numbers Are Wild

6 months of building at incredible speed

0
code changes in 6 months
0
changes EVERY DAY
0
lines of code (~4 novels)
0
software releases

130 people contributed to the project — but one person did the lion's share.

One Man Army

Mario wrote 80% of everything

2,350
of 2,920 changes
125
in one day (record)
  • He codes 7 days a week (no weekends off!)
  • His peak hour: midnight to 1am
  • He uses AI to help him code
  • And he's building an AI coder — it's recursive!

Growth Story

Code changes per month — watch the explosion

85
Aug
53
Sep
129
Oct
280
Nov
872
Dec
1,224
Jan

From 85 changes in August to 1,224 in January — a 14x increase as the project went viral.

The AI Building AI Pattern

This is where it gets mind-bending

🔄
Mario uses Pi (his own AI assistant) to build Pi. That's like a chef using their own recipe to make a better recipe book.
🤖
The Slack bot "mom" makes code changes on its own — 18 contributions and counting. The project even has a file (AGENTS.md) that teaches AI agents how to work on the code.

It's like a robot that builds better versions of itself!

How They Let People Help

The contribution system is like a club

1
Want to help? Raise your hand (open an issue)
2
A member vouches for you ("lgtm" = looks good to me)
3
A robot adds you to the guest list
4
Now you can contribute!
5
No guest list? Your contribution is automatically returned

Currently 118 people are on the guest list.

Ship Fast, Fix Fast

Mario's development philosophy

  • Release early, release often (2x per day!)
  • Fix bugs faster than they appear (488 bug fixes vs 318 new features)
  • Keep the core simple (only 4 basic tools)
  • Let extensions handle the fancy stuff
  • If you don't need it, don't build it
488
bug fixes
318
new features

Lessons for Any Team

Even non-technical teams can learn from this

1. Keep your core product simple

2. Let others extend — don't try to build everything

3. Automate repetitive tasks (bots for code review, contributor management)

4. Ship small updates frequently (better than big scary releases)

5. Document for both humans AND AI

6. Use AI to boost your productivity

7. Build in public — the community will help

The Viral Moment Timeline

Aug 2025 Mario starts building Pi quietly
Nov 2025 Peter builds OpenClaw using Pi as the engine
Late Jan 2026 OpenClaw goes viral — 145K GitHub stars in a week
Jan 2026 Anthropic asks for name change (trademark issue with "Clawd")
4 days later Renamed to Moltbot, then OpenClaw — community helps fast
Feb 2026 Featured on Syntax FM podcast, MIT Technology Review
Now Still growing — community of 130+ contributors

What This Means for the Future

👤
One person with AI can build what used to take a whole team. Mario proved it.
🔓
Open source AI tools are catching up to big tech. The best tools may be free ones.
🔄
The "AI building AI" pattern is accelerating development exponentially.
🏠
Local, privacy-respecting AI assistants are the trend. Your data stays yours.

Pi proves that simplicity wins —
even in the age of AI

agent37.club

agent37.club