Huan Li

Huan Li Follow

https://github.com/huan

Founding Human & Architect | Author @ Wechaty, Silicon Adventist, Serial Entrepreneur, Angel Investor, BurnerđŸ”„

Huan Li is the creator of Wechaty, and a leading voice in the software agent space. He is widely recognized for his contributions to conversational AI, cloud-native architecture, and the open-source ecosystem.

Huan is a GitHub Star, and the co-author of Chatbot 0 to 1 and Concise Handbook of TensorFlow 2. His work has earned accolades from Microsoft, Google, and Tencent—recognized as a Microsoft MVP, Google Developer Expert (GDE), and Tencent Cloud TVP.

As an AI architect and open-source evangelist, Huan frequently speaks at global tech conferences, sharing insights on building intelligent systems with CQRS, event-driven design, and LLM-powered agents.

Connect with him on LinkedIn or explore his open-source work on GitHub.


Entrepreneur & Angel Investor

Beyond engineering, Huan is a serial entrepreneur with deep roots in China’s early internet era. In 2001, he founded AKA.cn, delivering open-source enterprise solutions—including OEM hardware products like anti-spam email gateways and firewalls that later went public.

In 2005, he joined the founding team of YouKu.com (the “YouTube of China”) as CTO, and in 2007 launched JiWai.com, one of the earliest Twitter-like platforms integrated with QQ.

Since 2011, Huan has been an active angel investor, independently backing 100+ startups with a strong focus on developer tools, AI, and early-stage innovation.


Talks & Workshops

  1. Sessionize Machine Learning in JavaScript/TypeScript :Harness the power of ML in the versatile environment of JavaScript.
  2. Event-driven Programming with CQRS Wechaty
    Leveraging Event-driven Architecture in Wechaty with CQRS, enhancing system decoupling, microservices compatibility, and API integration through the wechaty-cqrs module.

Contact Me

Posts by Huan Li

From Message Chaos to Order: How I Rebuilt the Mailbox Library for XState@5

Lost messages are silent failures. Here’s why the actor mailbox pattern matters, how XState v5 enables it, and what changed in my Mailbox rewrite.

In engineering, Jan 08, 2026

Why LLMs Can Copy UUIDs (and How That Makes Agents Safer)

LLMs are stochastic. UUIDs are unforgiving. Yet models copy random IDs flawlessly—so let’s steal that trick and use it to build more robust Goodwin Checks for PromptWareOS.

In ideas, Dec 29, 2025

We Built Wooden Headphones: How We Stopped Cosplaying Linux and Started Building an OS

How we discovered we were cargo culting Linux instead of building a real operating system—and what happened when we stopped carving wooden headphones and started building real planes.

In engineering, story, Dec 27, 2025

We Tried to Make Our Prompts Shorter. We Made Them Longer Instead. Somehow, That Made Them Better.

The origin story of STOP Protocol and Ring Zero Linguistics—how we discovered that optimizing natural language for LLMs means maximizing character density, not minimizing word count.

In engineering, story, Dec 26, 2025

PromptWare v0.5: Enforcing Cognitive Integrity with Software Physics

In our last update, we introduced the concept of PromptWare: treating the LLM prompt as a kernel and the underlying code as hardware.

In engineering, Dec 21, 2025

Belief vs Control: /proc and /sys as an Interface for Agents

PromptWareOS borrows Linux’s greatest interface trick—virtual filesystems—and fixes the parts the kernel community would never dare redesign today: belief vs control, /proc vs /sys, a...

In engineering, Dec 20, 2025

Ship.Fail Crew: The First PromptWareOS Distribution

PromptWareOS gives you the boot protocol and kernel. Crew is the userland distro you mount as $HOME—so any GitHub repo can boot into an AI co‑founder environment with one AGENTS.md.

In engineering, Dec 19, 2025

Booting Intelligence: The Architecture of PromptWar̊e ØS v0.2

PromptWar̊e ØS v0.2 tackles context drift with a kernel memory subsystem, JIT-linked skills, and bootloader-driven mounts.

In engineering, Dec 18, 2025

PromptWar̊e ØS & Prompt‑Driven Architecture (PDA)

An ACM/IEEE‑style industry research report for AI co‑founder builders. Prompt‑Driven Architecture (PDA) is an emerging systems pattern where natural language prompts become the primar...

In engineering, Dec 17, 2025

Press0: Vibe Coding Your Way to a Physical Book

📖 Let’s be honest: writing a book is the easy part.

In announcement, Dec 16, 2025

Why Your AI “Two-Week Plan” Ships in 30 Minutes

A practical specification for translating between human wall-clock time and agent-subjective time through the Clock Bridge

In ideas, Dec 14, 2025

Shipping PPL: A License for the Age of PromptWar̊e ØS

We needed a license for our AI co-founders. So we built one. Introducing the Public Prompt License (PPL) v0.1.

In announcement, Dec 13, 2025

PromptWar̊e ØS Ships: Unix Architecture for Your AI Co-Founders

Stop copy-pasting prompts. Start booting an OS. PromptWar̊e ØS applies Unix architecture principles to transform scattered prompts into a structured, distributed, and bootable environ...

In announcement, Dec 13, 2025

PromptWar̊e ØS: One Line to Boot Your AI Co‑Founders

Turn fragile, copy‑pasted prompts into a shared operating system for your AI co‑founders, booted by a single line in their setup.

In ideas, Dec 08, 2025

When English Hits Ring 0: A Field Guide to PromptWar̊e ØS

Software 1.0 was code, 2.0 was data, 3.0 is prompts. Who owns prompts, how do we license them, and what happens when English effectively runs at ring 0?

In engineering, Dec 03, 2025

The Missing Feature in AI Coding Agents — And Why Only Claude Code Delivers It

A Deep Dive into Branching Conversations, Editable Prompts, and the Future of Terminal AI

In article, Dec 01, 2025

The Vibe Coding Cloud Protocol: One Tree to Rule Your Infrastructure

I used to dread opening my Azure portal—it was a graveyard of forgotten resources and confusing bills. Today, it’s a clean machine that matches exactly how I run my business. Here is ...

In engineering, Nov 29, 2025

Goodwin Checks: What Source Code Taught Me About Cognitive Health for Coding Agents

Before every mission, a woman asks a soldier a poker question. That tiny scene is the best design pattern I’ve seen for long-running AI agents.

In ideas, Nov 27, 2025

The Antigravity Effect: When Coding Becomes Orchestration

As an internal tester for Google Antigravity, I’ve had a front-row seat to the single biggest step-change in software development since the original IDE. But this isn’t just about f...

In tools, Nov 25, 2025

Confident-Cluster Decoding: A New Way to Make LLMs Less Random and More Trustworthy

Greedy decoding is too brittle, sampling is too random — Confident-Cluster Decoding (CCD) keeps only the tokens the model is truly confident about, exposes all top contenders, and tur...

In ideas, Nov 17, 2025

Announcing FireGen: Turn Firebase RTDB into Your Universal Generative AI API

FireGen is a Firebase Extension that transforms your Realtime Database into a universal Generative AI job queue, simplifying AI integration for solo founders and developers.

In announcement, Nov 13, 2025

Rethinking IAM: Designing a Human-Centered Command Line for Cloud Identity

A deep redesign of Google Cloud IAM’s language and CLI—simplified from first principles to minimize cognitive load and make identity management readable and intuitive.

In ideas, Nov 08, 2025

Rethinking Azure Billing: The Simple Design Microsoft Should Have Built

A founder’s deep dive into Azure’s confusing billing system — MCA, EA, MOSP, and beyond — and a redesign proposal that finally makes sense. This reimagined model simplifies Azure bill...

In ideas, Nov 04, 2025

Vibe Coding to Production: The Case for a Two-Stage Pipeline

Purpose: Present PreAngel’s current best-practice two-stage deployment model — powered by GitHub Actions, Cloud Run, Artifact Registry, and Workload Identity Federation (WIF). This ve...

In guide, Oct 13, 2025

🧭 Decoding the Chaos: The Ultimate Guide to Software Environment & Release Naming

🚀 Confused by terms like Preview, Beta, and Staging? You're not alone. This post breaks down the chaos into three clear axes every developer should know - By Huan Li — a developer onc...

In guide, Oct 12, 2025

Google Cloud Authentication — The Divio-Style Field Guide for Senior Full‑Stack Devs

A practical, copy‑pastable guide that untangles identities, credentials, tokens, ADC, OAuth 2.0, API keys, service accounts, impersonation, and Workload Identity Federation — organise...

In guide, Sep 12, 2025

All Interfaces Are Becoming Conversational

Huan's Law: 'Any app that can be talked to, will be talked to.' Explore the future of voice-first and conversational apps, where talking to technology becomes the norm.

In announcement, Apr 15, 2025