In April 2013, I was standing in a banquet hall in Wafangdian, a small city in northeast China, organizing a sales rally for 240 beer salespeople. I was twenty-eight. My job was making sure the LED screen worked, the banners were hung straight, and the team awards were printed on KT boards without typos. I had a clipboard. I had a seating chart for twenty-four tables. I had been at the company for five years and my title was HR Manager, but that afternoon, I was the guy making sure the bus drivers got paid 900 yuan each.
Last week, I submitted an iOS app to the App Store at 11pm, alone, from my apartment. Three AI agents were running in the background — one posting on X, one formatting content in Discord, one reading my Obsidian vault. I built the app in four days. I hadn't hired anyone.
Between the banquet hall and the apartment, there are twenty years. This is what happened in them.
The Beer Years
I joined Anheuser-Busch in Harbin in March 2008 as an HR specialist. Six months later, AB InBev completed its acquisition, and I was absorbed into the largest beer company on the planet. I was twenty-three. I knew nothing.
What I learned in the next eleven years was not HR. It was systems.
Every brewery acquisition followed a playbook — a literal binder of checklists for how to integrate a local Chinese brewery into a Belgian-Brazilian multinational without the local talent walking out the door. I became the person who ran those checklists.
In 2009, I was part of a team that laid off a fifth of the northern division's workforce. I was twenty-four. Over the next six years, I managed the people integration for six brewery acquisitions across China. I remember one factory manager in Shandong — leathery hands, cigarette going cold between his fingers — staring at the new pay band table I'd laid on his desk. He'd built that brewery from a two-line operation. He didn't argue. He just read every row, slowly, and then asked if I wanted tea. That silence taught me more about change management than any training module. What those years actually gave me was a skill I didn't know I was building: the ability to take a messy human system, find its structure, and make that structure legible to people who didn't design it.
By 2014, they made me Integration VP — the youngest in Asia Pacific. By 2016, I was running HR for Central China — five provinces, direct report to the division president. Corner office in Wuhan, a whiteboard covered in org charts I redrew every quarter, a driver who knew which highway exit to take for every brewery in the region. I was thirty-two, and by every measure the company used, I was succeeding.
But a system you master eventually becomes a system that masters you.
The Error Report
In September 2018, three months before I left AB InBev, I wrote a self-criticism report to my division president. Three incidents where I'd failed — a mishandled career conversation, a task executed on a guess instead of asking for clarification, and a situation I should have stopped but didn't.
The mistakes weren't the point. The pattern was. These weren't the errors of someone climbing. They were the errors of someone who had stopped learning inside a system he'd already mapped. The self-criticism was honest, but the honest conclusion — the one I didn't write in the report — was that I needed to leave.
I left in December 2018. Eleven years, one company, and a filing cabinet full of playbooks that had nowhere left to go. I cleaned out my desk on a Friday. Nobody threw a party.
Velocity
What happened next was fast.
I joined a vape startup called SnowPlus as HRBP Head — core founding team. At AB InBev, you optimize a system that already exists. At a startup, you build the system while the building is on fire.
New office. Empty desks. Whiteboards still wrapped in plastic. We hired. We shipped. Revenue went vertical. Retail stores opened faster than we could print employee handbooks. Then the government banned online sales of e-cigarettes. Overnight, half the business model evaporated. Then COVID hit. I restructured the company twice in the same year — once to survive the ban, once to survive the lockdown. Same conference room, different spreadsheet, same question: who stays.
SnowPlus taught me one thing that eleven years at AB InBev never could: the methodologies I'd accumulated — the integration playbooks, the budgeting systems, the restructuring muscle — weren't beer knowledge. They were portable operating systems. They worked at a vape startup the same way they worked at a brewery.
The methods were mine. The company had just been the training ground.
The Peak That Wasn't
In January 2021, I joined Longfor Group's rental and brokerage business as HR Senior Director. Within six months, I was promoted to HR General Manager. By July 2022, I was named Preparatory Partner — one of Longfor's highest honors, roughly equivalent to being flagged as future senior leadership.
Two promotions in one year. Equity. A team that filled a floor. I remember the afternoon the Partner announcement went out — my phone buzzing with congratulations while I sat in a glass-walled office watching the Shenzhen skyline turn orange. That view was supposed to feel like arrival. It felt like a question mark.
What exactly was I building?
At AB InBev, I was building integration playbooks. At SnowPlus, I was building an organization from zero. At Longfor, I was optimizing someone else's machine. Making it run faster, run leaner, run with fewer breakdowns. But the machine wasn't mine. The strategy wasn't mine. The product wasn't mine.
In May 2024, I left. My contract had just been renewed three months earlier.
The Inversion
I don't have a clean narrative for how I got from Longfor to running three AI agents from my apartment. There was no eureka moment. There was an MBA.
I entered Peking University's Guanghua School of Management in September 2023, while still at Longfor. Fifty-three courses over two years. Strategy, corporate finance, organizational behavior, marketing, leadership — the full canonical MBA curriculum, plus electives in entrepreneurship and industrial analysis.
What I didn't expect was what the courses would become.
After I left Longfor, I had time. And I had a habit — twenty years of compulsive methodological extraction. At AB InBev, every project generated a playbook. At SnowPlus, every crisis generated a post-mortem. At Longfor, every promotion cycle generated a self-assessment framework. I don't just go through experiences. I extract the transferable structure and write it down.
So I did the same thing with 53 MBA courses. I built a six-agent pipeline inside Claude Code — a data fetcher, three parallel distillers, an auditor, and me as coordinator. The pipeline read my Notion database of lecture transcripts, PDFs, and discussion notes, and produced reusable Concepts, Skills, distilled session notes, Maps of Content. Two weeks.
The first time I ran it end-to-end, I was sitting at my kitchen table at 2am. The terminal scrolled — agents reading transcripts, extracting frameworks, cross-referencing for duplicates, filing into the vault. I didn't touch the keyboard for forty minutes. When it stopped, I opened Obsidian and searched "authority." There it was: Weber's Three Types of Authority, fully structured, linked to three other Concepts I hadn't thought to connect. Two semesters of lectures, distilled into something I could use in a conversation tomorrow morning.
That pipeline did not come from learning AI. It came from twenty years of doing exactly the same thing by hand — extract, structure, deduplicate, link — and suddenly having a tool that could do the mechanical part at scale.
The Compound
In November 2024, I co-founded WHoos Solutions — an AI-first consulting and tech services company, incubated at the Peking University Science Park. Manufacturing, supply chain simulation, emergency management — real projects, real clients, first year.
Simultaneously, I shipped three iOS apps in thirty days. One Page, a journaling app with invisible AI reflections. FileFlow, rule-based file organization. MARGIN, a schedule inbox that went from idea to App Store in four days.
Simultaneously, I built a content operation that runs on three AI agents, an Obsidian vault, and a terminal. Every morning I open the laptop, check what the bots drafted overnight, approve or kill each piece, and close the lid. Daily posts, weekly threads, monthly articles — all in English, all sourced from Chinese notes, all passing through a human review gate before anything goes live.
None of these are separate stories. They're the same story.
The methodological extraction habit from AB InBev became the knowledge distillery. The org-building speed from SnowPlus became the product shipping speed. The data-driven efficiency obsession from Longfor became the cost discipline in my 15-rule product kill list. The integration playbooks became Claude Code skills.
Everything I spent twenty years learning how to do inside other people's companies, I now do inside my own system. The difference is that the system amplifies one person instead of optimizing one thousand.
What I Don't Know
I don't know if WHoos Solutions will exist in three years. I don't know if the iOS apps will find an audience beyond the few hundred people who need exactly what they do. I don't know if running three AI bots from a terminal is the future of content operations or just a weird intermediate stage before something better replaces it.
I don't know if an MBA from Peking University matters more or less than the eleven years I spent learning how to tell a factory manager that his job grade is changing.
I don't know if I left Longfor too late or too early. Some days it feels like both.
What I know is that every skill I built in twenty years of working for other people turned out to be a component in something I couldn't have designed in advance. The playbook habit became a pipeline. The speed became shipping velocity. The efficiency obsession became a product philosophy. The self-criticism reports became a quality audit framework.
Nothing was wasted.
But nothing was planned either.
The Shortest Version
From beer salesman's HR guy to one-person AI company. Twenty years, five industries, six companies, two promotions in one year, one departure, fifty-three courses, three apps, three bots, one vault.
The system runs. I'm the only employee.
I'm @UncleJAI. This is the third piece in a series about building alone with AI. The first was about the architecture. The second was about the shared memory layer. This one is about the person who needed twenty years to be ready for both.
