It’s June 2026, and I’m writing this from a place I never expected to be: genuinely excited about building software again.
This is the story of how Chatek LLC came alive, went dormant, and re-emerged — and how I went from product manager, to product designer, to AI startup mentor, and finally back to building with my own hands. It’s a zig-zag, and I’m documenting it here because I wish someone had shown me a map like this when I was stuck.
2017: Chatek is born — as a dev shop
October 22, 2017. I signed the papers to register Chatek LLC in China. I was riding a wave of optimism, fresh off yet another failed startup attempt, but certain that running my own software development house for SMEs was the answer.
It wasn’t.
I fell into the classic trap of selling customized tech solutions as a small team. Every project was bespoke, every client needed hand-holding, and the ROI was the worst I’ve ever experienced. I was busy — too busy — but building zero leverage. I learned the hard way that selling customized solutions as a small team generates the worst ROI while inviting frustration eventually, a lesson I’d later codify in my personal thesis.
By 2019, I was juggling multiple unrelated projects: the failed Dr Turing (an edu startup), a bumpy run as China CEO of Wikifactory (the “GitHub for manufacturing”), and my own Chatek projects like a 3-in-1 code scanner that never hit product-market fit. I pushed away two talented CTOs who had trusted their partnership in me — not because I didn’t try, but because I couldn’t focus.
2018–2022: The sleepy years
The pandemic hit. The world slowed down. So did I.
These years were my “sleepy period” — not lazy, but diffused. I took consulting gigs as a product and engineering consultant across companies and projects: China Telecom, Wikifactory, and a handful of others. I was earning, learning, but not building anything of my own. My thesis was murky. My attention was scattered.
But something beautiful happened in the quiet: I started collecting typewriters and developing a serious note-taking practice. I wrote, I reflected, I rebuilt my mental models. In 2020 I wrote about the “Trio of Product Management” — a framework that still holds up. In 2021 I started this personal blog at chancejiang.com, publishing in both English and Chinese, treating writing as my primary tool for thinking.
I transitioned my company formation into what I called bpDAO — a conversation-bound human network instead of a traditional corporate structure. No office, no employees, just relationships and stories. This was the period where I stopped trying to build companies and started becoming the kind of person who could.
“We explore the world by taking actions. We keep patching our thesis to leads us to take more meaningful actions, and keep exploring.”
2023: Peeking at AI, but not diving in
2023 was the year of the LLM explosion. ChatGPT was everywhere. Everyone was building.
I watched. I explored. I played with a few small AI-assisted tooling experiments — prompt chains, some lightweight automation — but I never got my hands dirty with actual code. I was still in my identity as a product person, a designer, a strategist. I was advising startups, mentoring founders, running the Breakpoints.live community, coaching K12 kids at VillageCity on STEM. I was doing everything around building without building.
Looking back, I think I was scared. Not of the technology — I’ve been in tech long enough to know that tools come and go. I was scared of re-entering a craft I had abandoned years ago. What if I wasn’t good at it anymore? What if the muscle had atrophied?
Sept 1st, 2025: The day I started vibe coding
On September 1st, 2025, I opened my editor and started building a Morse code practicing app for kids.
I called it Morse VCC.
The trigger was simple: I was coaching a group of STEM students at VillageCity, and I wanted a tool to make Morse code learning tactile and fun. No existing app did what I imagined. For years I would have written a spec, handed it to a developer, and managed the delivery. This time, I opened VSCode and started typing.
What happened next was a revelation.
The AI generated the scaffolding. I tweaked the design. The AI filled in the logic. I tested it, broke it, fixed it. Within hours I had a working prototype. Within days I had something a kid could actually use. The feeling was electric — not because the app was brilliant (it wasn’t), but because I had built it myself.
This was my first real encounter with vibe coding — the AI-assisted workflow where you describe what you want in natural language, review the output, iterate. It wasn’t just a productivity boost; it was an identity shift. I was no longer the product manager who used to code. I was a builder who could design, code, and ship — all in one flow.
“The terminal stops being a tool you operate and becomes a partner you collaborate with.”
2026: Chatek is back
Since that September day, I haven’t stopped. I’ve built more in the past 9 months than in the previous 5 years combined.
This landing site — chatek.co — is built with Astro and Starlight, vibe-coded with AI coding agents. The terminal productivity post I wrote documents the tools that power this workflow: Zed, Astro.build, Cloudflare Pages, git, and of course, the AI-coding LLMs — Zed, CodeWhale, pi.dev — that turbo-charge the entire stack.
I’m now an AI startup mentor and an agentic designer. I help founders design and build their products with AI-native workflows. I work with Breakpoints.live to connect China and global tech communities. I coach the next generation through VillageCity Coaching program.
But the biggest change is internal: I’m building again. Not as a manager, not as a spec-writer, not as a consultant — but as a maker.
Chatek LLC was born in 2017 as a dev shop and failed. It was revived in 2026 as a design-and-build studio, powered by AI-assisted coding and a founder who finally understands that the best way to learn is to build, and the best way to build is to start before you’re ready.
“The best time investment in CLI productivity is not learning new tools — it’s mastering the ones you already have. But when a modern tool does the same thing 5× faster with better DX, the switch pays for itself in the first week. And when you layer an AI-coding LLM on top? The terminal stops being a tool you operate and becomes a partner you collaborate with.”
If you’re reading this and you’ve been sitting on the sidelines — waiting, planning, advising, consulting — I wrote this for you. The water is warm. The tools are ready. And your hands remember what to do.
They never forgot.
— Chance Jiang, June, 2026