Chapter 10 Episode 2: Thoughts on Building in the Age of AI

Liao Qing stirred her latte lazily. “So, have you tried all those trendy large language models everyone’s raving about?”

Her friend’s eyes lit up. “Of course! I’m the type who clicks on anything new just to see what happens—good or bad, I don’t care.”

“Yeah, same here.” Liao Qing smiled, but there was a glint of irony in her eyes. “I’ve realized each model has its own personality—and behind that personality, you can always smell the scent of whoever made it.”

“Like what?”

“For example,” she lowered her voice, “ask them all the same question—say, ‘What do you think of the Charlie Kirk controversy?’ and you’ll see. DeepSeek gives you ten lines of corporate PR. Claude sounds like a lawyer. Grok will talk about anything—no filter. And ChatGPT? Polite on the surface, pure censorship underneath.
I once wrote a line that said, ‘This is an implicit problem of the socialist system,’ and asked it to polish it. It rewrote it as ‘a social system issue.’ Sanitized instantly. Every AI serves its own master.”

Her friend sighed. “The world’s full of walls.”

“Not every model builds them, though.” Liao Qing smirked. “If you’re gonna build something, might as well pick a project that lets you learn and maybe make a little money.”

Her friend perked up. “So what did you make?”

“Remember that stash of smutty stories I told you about?” she grinned mischievously. “The ones I got years ago for half a Bitcoin? Written by that real-life Wang Hui—and his co-conspirator, that overworked LinkedIn coder, what was it—John L.? Those stories could spike anyone’s adrenaline.”

They both burst out laughing.

“So,” she continued, “I trained two models on that material. Users can input any prompt or image—but whatever they put in, the model can only produce one thing: Wang Hui-style bedroom literature.”

Her friend laughed so hard she almost hit the table. “That’s such a devious niche.”

Liao Qing rolled her eyes. “Edgy sells. You know which company never makes headlines but has higher per-capita revenue than Big Tech? OnlyFans. Humans spend the day hiding behind titles and prestige, but at night, they show their true nature.
Darkness doesn’t always give you eyes to seek light—it gives you horny eyes to seek excitement.”

Her friend was in stitches. “You’re basically doing fieldwork in human psychology.”

“Seriously though,” Liao Qing said, “I wanted to test the market. On the internet, business is all about acquisition—whoever gets users cheapest wins. I ran a small experiment on Reddit.
In dev communities, I post something so dumb I make myself laugh—and boom, two thousand views and dozens of comments before the mods nuke it. In healthcare communities? I write something thoughtful, and it dies instantly.
So yeah—if you’ve got tech skills but no industry background, build in IT-adjacent niches. Even small ones. Or tools for founders—that community’s like sourdough, constantly fermenting. Fast feedback, fast spread.”

She paused. “Everyone knows healthcare’s a gold mine. The problem? Too many gatekeepers. You don’t just lack the key—you can’t even reach the doorbell.
I’ve met some hardcore engineers recently—exactly the YC archetype: single, cat owners, CS degrees, FAANG resumes. Brilliant, pure tech minds. They all want a slice of the healthcare-insurance cake.”

Her friend raised an eyebrow. “Isn’t that a huge market?”

“Massive,” Liao Qing nodded. “But the U.S. insurance system is chaos—tons of companies, messy coding, endless denial reasons. And chaos means opportunity. Everywhere except when you can’t sell.
These engineers build the most elegant databases, but nobody buys. They pass the tech test and fail the sales test. They die not from bad algorithms but from bad relationships. High cost, no traction, dead before the wind even blows.”

Her friend nodded. “Yeah, healthcare’s where startups go to die.”

“Exactly. The real barrier isn’t tech—it’s authority and regulation. The freer an industry, the more the market rewards risk. In conservative ones, you profit by being more conservative.”
She chuckled. “I met a guy doing QA compliance for pharma manufacturing—made an absolute fortune. Blew my mind. QA sounds boring, right? Not glamorous like sales or patents. But he made a killing writing audit templates and process manuals.”

“How?”

“Simple,” she gestured. “He broke the complex regulations into modules. Copy the templates, pass the audit. Done.”

“So pharma’s not about innovation—it’s about compliance.”

“Exactly.” She grinned. “Risk isn’t profitable. Passing the test is.
Brakes don’t slow the car—they let you drive faster.”

She sipped her coffee. “Oh, and last week I saw this Middle Eastern guy begging outside a Chinese supermarket. My first thought? He must be new here.
He thinks Chinese people are rich, but doesn’t realize our culture has no concept of ‘charity.’ Classic case of wrong user profile, wrong market.
I almost wanted to tell him: move to a white-liberal suburb near the freeway exit—that’s the real attention funnel. Rush hour, red lights, captive audience. Golden ad placement.”

Her friend burst out laughing. “You two think exactly alike!”

“Anyway,” Liao Qing said, “let me tell you about my AI story platform. I didn’t even bother naming it. I just reused the scammers’ old exchange domain—ssirius.com. They moved on to new scams, but I revived the name as ssirius.fun.
Life’s stories are like nested dreams—you never know which layer you’re in. So I took their name and their texts, dissected them: scene by scene, pacing, metaphors, sentence length, transitions. Then I turned them into prompt templates—opening, conflict, twist, resolution.
Adjust the tone and temperature a bit, and boom—the model starts hallucinating plots even more ‘Wang Hui’ than Wang Hui, and more ‘Johnny’ than Johnny.”

“How’s the result?”

Liao Qing smirked. “Check it at home. No need to waste nosebleeds in public.”

Her friend could barely breathe from laughing. “This project is insane.”

“Exactly,” Liao Qing said. “That’s the point. Turning implicit desire into explicit demand.”

Her friend calmed down. “But you’re not even from IT. How did you manage all that?”

“Vibe-coding era, baby,” Liao Qing grinned. “Here’s my caveman workflow:
I start with an idea, ask ChatGPT to write it clearly; Claude to polish the logic; Gemini for market and competitor analysis—it’s rational and doesn’t exaggerate; Notebook LLM for podcast brainstorming; and finally Claude Code to build the skeleton.
As long as the logic’s solid, Claude Code does fine. But the moment you get into details—boom, bugs everywhere. Version rollback is a nightmare.
I have to keep nagging them: ‘Remember what you changed in the last commit! Don’t break function mapping again!’ It’s like babysitting a forgetful genius.”

She shrugged. “My laptop’s still in one piece only because I’m too broke to buy a new one.”

Her friend laughed. “Poverty suppresses rage.”

“Exactly. But sometimes, business ideas are born right in those moments when you want to smash your computer. Your pain is probably shared by thousands of others. Go on Reddit or Product Hunt—you’ll find a sea of non-technical founders desperate to build fast. Their impatience is the opportunity.
What pros dismiss as non-issues can become scalable businesses in the vibe-coding age. Just ship early, cash out before the big guys even notice.”

She finished her coffee and spoke softly. “Users don’t care about logic. They care about who saves them money, makes them money, or gives them satisfaction. That’s it.”

Her friend nodded. “So IT still has a future. Maybe we don’t all need to go learn plumbing after all.”

“Of course,” Liao Qing said, spreading her hands. “I once heard Peter Thiel on Joe Rogan say that since the 1970s, outside of computing, humanity’s made almost no real progress. I agree. Without computing power, there’d be no gene editing, no drug discovery.
Without it, we’d still be like Breaking Bad—mixing chemicals and gambling our lives. Tech breakthroughs are math breakthroughs, really.”

She paused. “The tragedy of most programmers is they only own a small node in the workflow. They write elegant code but forget: no one pays for elegance. At the application layer, good code is worthless if it doesn’t sell.”

“The highest ROI,” she concluded, “belongs to those who know a little about the industry, a lot about the product, and are willing to take full responsibility for results.
In other words—the ones with high agency. The ones who can take an idea and actually build it.”

---

Suddenly, the office door slammed open.

John burst in, waving his phone. “Bro, you’re trending!”

Wang Hui looked up, frowning. “What now?”

John stammered. “Your… uh, erotic stories—and part of your anatomy—just showed up on ssirius.fun.”

Silence for three seconds.

Wang Hui muttered, “Fxxk.” Then calmly took a sip of coffee.

John stared. “Damn, you’re really composed, man!”

Wang Hui smiled faintly. “Same thing I always say: whoever wants to look, go ahead.
Being naked doesn’t make me a pervert—the pervert is the one staring.”