On Entrepreneurship, Taste, and Choosing What Not to Do

Weekly Health Log Summary:

On Entrepreneurship

Rousseau wrote: “Man is born free, yet everywhere he is in chains.”

After six years of founding, I feel that truth more deeply than ever.

Many people start businesses seeking freedom—but along the way, many end up wearing more chains: chained to growth targets, fundraising milestones, performance anxiety, or even their own ambition.

True freedom may not mean “doing whatever you want.” It means having real choice—and the capacity to act on it—within clear, self-defined boundaries.

Yesterday, Tony sent me a thoughtful message on WeChat:

Run your business to maximize profit—not size.
Start a company for freedom, health, and joy—not for the sake of starting one.
Never drag yourself or your family into hardship just to “build something.”
Once you reach a certain scale, practice restraint—put safety first.
In entrepreneurship, choice outweighs effort. Compete through differentiation—not scale.
A good business is one where others can’t figure out how you make money—not one where you flaunt your revenue.
Real opportunity lives in anti-consensus: ignore trends, build moats instead of chasing growth, think in cycles—not quarters.
When you find something others take three days to solve—and you nail in three minutes—that’s your key to real wealth.

Underlying all this is a single idea: entrepreneurial success doesn’t require “getting big.” It requires finding a sustainable, deeply personal way of living—one that aligns with who you are and what you value.

I joked back: “Every sentence here sounds like it was written about me.”

Most founders share a core trait: they want to write the rules themselves.
We’re not afraid of hard work—we’re unwilling to hand over our time, direction, or destiny to someone else.

When launching our AI Leadership course, Yangyang and I decided early on to “keep it light”: a business we could run while fishing and running—no funding, no heavy infrastructure, no rush. Many didn’t get it. But we knew: we weren’t optimizing for maximum profit. We were optimizing for maximum freedom.
That’s a deliberate refusal to be hijacked—by funding opportunities, market hype, or capital-driven timelines—away from our original purpose.

That’s why our current AI project thinking revolves around recurring keywords: Think Small, micro-case studies, AI First, and “AI employees.”
We want to build an AI-powered business system that’s lean, responsive, and—above all—unbound.

On Taste

Cursor is currently the highest-valued AI coding tool—and I use it daily for prompt engineering and lightweight software design.

In a recent interview, CEO Michael Truell shared insights on AI programming and Cursor’s evolution. A central theme? The irreplaceable role of taste.

He argues that “taste”—not technical fluency—is the last human stronghold in software development. Whether judging interface aesthetics or architectural logic, AI can automate how—but humans must still decide what and why well.

He calls today’s coding “human compilation”: You know exactly what you want—but must painstakingly translate it into loops, conditionals, and variables. In the future, AI will handle the translation. But the original intent—the “what”—remains yours to define.

This resonates strongly with our AI Leadership course module on aesthetic literacy and experiential literacy.

1. Taste goes far beyond visual design
   Yes, it includes color, typography, and layout—but also product logic, user flow, and feature restraint. Is the interaction intuitive? Does the feature set feel just right—neither bloated nor barren? That judgment is taste.

2. Programming is fundamentally “human compilation”
   You hold a mental model; then you manually decompose it into machine-executable steps. As AI absorbs more implementation detail, our energy shifts upstream—to the essential question: What do I truly want?

3. Taste cannot be automated
   Even if AI writes flawless code, taste remains human-exclusive. It draws on deep experience, contextual intuition, and empathetic understanding of users’ unspoken needs. AI executes. Humans intend—and refine intention.

4. The developer’s role is evolving
   From “coder” to logic designer or product aesthetician. As technical barriers fall, the differentiator becomes taste: Can you conceive and shape something genuinely valuable, usable, and beautiful?

5. Taste shapes teams—and products
   Early hires define a company’s taste ceiling. Cursor’s first 10 engineers weren’t just coders—they had product sense, commercial awareness, and cross-domain fluency. That taste-first hiring culture is why they ship great tools so fast.

Interview source: youtube.com

What to Do—and What Not To Do

A colleague asked: How do you decide what’s actionable now—and what to ignore?

I joked: “Chairman Mao already gave us the answer: Focus on the principal contradiction.”

He replied: “Great principle—but how do you actually identify it?”

I paused, then said:
First, know what you want.
Second, map the critical path—the few essential elements that make that outcome possible.

Knowing what you want is often the harder part.
But even without a long-term vision, short-term goals are usually within reach.

Still, that’s abstract—so I walked him through two concrete examples. In practice, this is just Critical Path Method applied to daily work.

Later, I added:
Yes, we’re bombarded by information—and our own swirling thoughts.
But with hindsight, most ideas turn out not worth doing.
What matters is disciplined alignment: every action, every experiment, every iteration should orbit your core direction and primary goal.
So clarity on what you truly want isn’t nice to have—it’s non-negotiable.

Managing Ideas

1. To tame idea overload, rely on the 80/20 rule and Occam’s Razor—shave away the unnecessary, regularly.
2. When everything looks like an opportunity, pause: Most of those “opportunities” aren’t yours. Stay curious—but don’t feel compelled to chase them all. Focus remains a strategic superpower.
3. The best strategy isn’t hunting more opportunities—it’s finding your domain and building lasting advantage there.
4. More important than managing ideas is managing outcomes. Every action yields a result—but only some results deserve sustained investment.
5. Which ideas merit action depends on what matters most right now—and, again, what you truly want.
6. At its root, idea management is about building a dynamic balance system: one that supports both divergence (for creativity) and convergence (for execution), innovation and accumulation.

AI Systems and Training Models

A friend attended an in-person AI marketing workshop and shared screenshots with me.

The training centered on two integrated systems:

• An AI Marketing System, bundling AI-powered short-video acquisition: auto-scraping competitor topics, AI scriptwriting, AI-assisted remixing, AI-driven repurposing, and knowledge-base management.
• An AI-SCRM System, embedding AI into customer relationship management—especially AI-augmented sales scripting, built on AI + domain-specific knowledge bases.

Individually, each capability exists as standalone (free or paid) tools. But for traditional SMEs, stitching them together is prohibitively complex. This company’s strength? Integration. They’ve unified what customers actually care about.

My friend was intrigued—until he learned the pricing model:
Not a system license. Instead: a 3-day, 2-night in-person workshop + 1-year access to the AI systems (for up to 20 accounts) + separate billing for AI compute.
Total: ¥100,000 per person.

I responded: “That model actually makes sense. Selling just the software would yield thin margins—and without methodology, the system’s potential stays locked. Plus, pure SaaS—AI or not—is brutally hard in China. Most lose money. Wrap it in training, add AI branding, and suddenly the barrier rises.”

How to Find AI Project Ideas

Inspired by Y Combinator’s interview: How To Get AI Startup Ideas
Source: youtube.com

This talk offers grounded, actionable advice—not hype—for spotting real AI opportunities:

1. Now is the best time to start
   Rapid AI progress has unlocked new possibilities across every traditional domain. YC urges technically skilled founders to begin—even without a polished idea. Re-examine old workflows with fresh AI eyes. Where does AI perform exceptionally well? Pair that with your domain insight.

1. Where do good ideas come from?
   Avoid hackathon-style gimmicks or trend-chasing. The strongest ideas emerge from:
   • Your own lived experience (e.g., Salient’s founder spotted loan-collection inefficiencies while at Tesla Finance), or
   • Deep immersion in a new industry (e.g., Diode’s founder fused hardware + software expertise to build an AI PCB collaboration tool).
     Other examples: Spur, DataCurve, David AI, Can of Soup, Happenstance—all rooted in founder-specific insight.
2. “Go out” or “look inward”
   Ideas grow from real problems—not speculation. Either:
   • Mine your past: What frustrated you? What did you automate instinctively?
   • Enter new terrain: Intern, shadow workers, or “go undercover” in unfamiliar roles. Ask: What repetitive tasks scream for AI? And crucially: If not me, who else could solve this?
3. Other practical notes
   • Talk to sharp people. Join cutting-edge teams. Proximity accelerates insight.
   • Don’t avoid crowded markets—if your solution truly works, you’ll stand out.
   • Be patient. Most successful startups pivoted multiple times before finding fit.
   • Prioritize repetition + outsourcing: prime targets for AI automation.
   • Observe friends, family, colleagues—their daily friction is your signal.
   • Build solutions, not “AI products.”
   • Either become a domain expert—or partner deeply with one.
   • Your best idea is likely hiding in what you already know—and do well.