📝 Articles

I stopped trying to build products that scale globally by default.

Not because demand disappeared.
But because the assumptions behind global had changed.

APIs are no longer neutral.
Data doesn’t move freely.
Dependencies that once accelerated you can now break you.

What looked like a product decision was actually a response to something bigger:
👉 We are no longer in an era of pure globalisation.
👉 We are in an era of constrained globalisation.

For years, the playbook was simple:
Build once. Scale everywhere.
Globalisation = horizontal integration at scale.

That world is receding.

What’s emerging instead:
Selective vertical integration (own what matters)
Regional horizontal integration (work with aligned partners)
Systems designed with constraints upfront

This led me to a simple framework I now think about constantly:
The 3Cs + 1
Compute (what infrastructure you can access)
Capital (who funds you—and what that implies)
Compliance (where you can operate)
Coordination (your ability to navigate all three)

Constraints are no longer edge cases.
They are design inputs.

That creates something I’d call: constraint arbitrage.

Not owning everything.
But navigating fragmentation better than others.

I wrote a deeper piece breaking this down into a full framework + strategy matrix -
https://lnkd.in/gZrvhDYW

#AI #ArtificialIntelligence #Philosophy #Innovation #ProductThinking #FutureOfWork #e27 #Startups #Cognition #PromptEngineering    

🧠 What if AI isn’t intelligence… but a mirror?
My latest piece just went live on e27 — and it started from a simple realization during a conversation: “What you prompt is what you get.”

At first glance, it sounds like a technical truth about AI.
But the deeper I explored it, the more it revealed something about us.

We often ask:
- Is AI intelligent?
- Can machines think?
- Will AI replace human cognition?

But maybe we’re asking the wrong question.

Because what AI really does is not think.

It reflects.

Through this lens, a few ideas emerged:
🔹 Intelligence may not be knowledge — but pattern recognition
Both humans and AI operate by detecting, organizing, and responding to patterns.
🔹 Creation is not from nothing — it is recombination
Originality isn’t magic. It’s navigating pattern space in ways others haven’t.
🔹 Prompting is thinking made visible
The quality of AI output is a direct reflection of the clarity of your own thought.
🔹 Human + AI is not collaboration — it’s recursion
You’re not just using AI.
You’re seeing your own cognition… reflected back at you.

And perhaps the most poetic realization:
Like Zhuangzi’s butterflies —
are we thinking with AI…
or is AI thinking through us?

This isn’t a technical article.
It’s an exploration of the non-technical dimensions of AI:
- consciousness
- pattern
- reflection
- and the nature of intelligence itself

If you’re building, writing, or thinking in the age of AI — this perspective might shift how you engage with it.

The e27 article is an abridged version of an original post:
https://lnkd.in/gAFrxnQz

👉 Read the full article on e27
https://lnkd.in/gRybHz7s

#AI #ArtificialIntelligence #Philosophy #Innovation #ProductThinking #FutureOfWork #e27 #Startups #Cognition #PromptEngineering    

Most AI upskilling efforts are treading water.
Not because people aren’t trying.
Not because the tools aren’t powerful.
    
But because we are ignoring something deeper.
I recently explored this in my article:“The grains of structural friction: Why AI upskilling is failing the power law” 

Here’s the core idea:
We assume skills scale linearly —learn a bit → improve a bit.

But the reality is not a linear relationship but a power law. This is expressed by differing outcomes — small number of individuals and organizations capturing disproportionate value from AI, while the majority gain average result. Until we understand it, we cannot address the gap which will continue to widen.
    
#AI #Upskilling #FutureOfWork #PowerLaw #Innovation #Leadership

For much of history, identity was built on competence:
What you knew.
What you produced.
Where you added value.

But when AI starts to write, code, analyze, and reason—that foundation of identity begins to disrupt.

And as AI entrenches, a question that is becoming harder to avoid surfaces:
Who are you, when what you produce is no longer uniquely yours?

In this piece, I explore:
• Why AI feels like identity loss but is not
• Why “Who am I?” becomes the one question no AI can answer

#AI #FutureOfWork #Philosophy #Identity #Technology    

Most organisations are still built like pyramids.

But work no longer behaves that way.

AI is quietly removing the need for coordination at scale—the very reason pyramids existed in the first place.

What replaces it isn’t flatter hierarchies.

It’s something more radical:
→ Individuals with the output of teams
→ Teams that assemble and dissolve on demand
→ Organisations that behave more like swarms than structures

We are not just changing how work gets done.
We are changing how work is organised.

In this piece, I explore:
- Why “team of one” is no longer a metaphor
- How capability is recomposing into fluid swarms
- Why advantage is shifting from scale → speed of reconfiguration
- And why places like Singapore may be uniquely positioned for this shift

This is not a future scenario.

It’s already happening.
    

Most people think AI fails because it’s wrong.

It doesn’t.

It fails because we overload it.


While building my API client, I hit a wall.

At first, AI felt incredibly powerful:
• Full UI generated
• Components wired together
• Everything in one go

Then things started breaking.

Fix one bug → two new bugs appear
Add a feature → something else disappears

A natural first instinct could be:
“Maybe the model isn’t good enough.”
“Maybe I need a better prompt.”


Thanks to my sociological background - I asked differently:
 “You seem to make more mistakes as the system grows. How can I help?”


DeepSeek was surprised by my question, and its answer also surprised me, and reframed everything: “I don’t get tired. But I do get crowded. Think of me as a desk, put too many papers on it, and things start falling off.”

    
The answer i was prompting was actually its real issue/constraint, and is this:
AI doesn’t run out of memory.
It runs out of attention.

When you push too much into one system, things start to fall apart.


Not because AI is flawed — but because the way we’re using it is.

The breakthrough was simple:

Instead of:
“Build everything”

I scoped to:
“Build this component” → test
“Add the next piece” → test
“Then integrate”

Everything changed.
Fewer bugs. Faster cycles. More control.

The shift:
The AI builds fast.
The human keeps the system coherent.

Sensing the limits is more like a balancing act.
Systems break when you try to engineer without understanding all systems have implicit constraints.

Building with AI is hence closer to a dance.

This article got published just at the right time — launching around July 4th a new API Client — looking to redefine the space.

Did ChatGPT Kill the Edu-Star?

Chegg was widely popular among students for the past two decades, and was also a Wall Street favourite. However, on May 2, 2023, Chegg became one of the first major companies to publicly admit that OpenAI's ChatGPT was directly hurting its subscriber growth. Following that Q1 earnings call, Chegg's stock plummeted nearly 50% overnight as students abandoned paid subscriptions for free AI chatbots.

Chegg didn’t fail because of poor execution.

Chegg didn’t fail because it stood still.

It failed because: The layer it operated on collapsed.

On hindsight, could Chegg have done better?
Likely not.

There was no precedent.
And the shift came fast — and unforgiving.

If your product can be replicated via a prompt, you are running against time.
We are entering a phase where competition doesn’t just improve products.

It erases categories.
Some businesses are Casualty Stacks while some are Value Fortresses

The difference is structural — not executional.

To make this tangible, I built a simple tool that helps diagnose the defensibility of your idea:

https://jiunnhwa.github.io/Value-Fortress-Stack/

It helps to shine a light on an uncomfortable question:

If AI attacked your business tomorrow, would it survive?

Build, but do not build the next Chegg, mate.

“Vibe coding” has made building feel accessible to everyone. But it hasn’t made outcomes equal. If anything, the gap is widening. Because AI doesn’t replace thinking—it reflects it. What you get depends on how you think. The more precise framing is this: AI has lowered the floor of app development, but raised the ceiling of what good looks like.

Most discussions about AI still revolve around one question: Will AI replace us? In my latest article, I argue that the real question may not be whether AI shall terminate humankind in the way Hollywood imagined… But whether it will quietly terminate the systems and notions that define our concept of human value today. The rise of AI represents a tectonic shift—not just capable of rewriting workflows and industries, but possibly reshaping the socio-economic assumptions and sub-strata that they are built on. And that puts us at a fork in the road: → One path leads to concentration—where AI-driven productivity is owned by the few, and the many become economically irrelevant. → The other leads to distribution—where humanity, liberated from labour, unlocks new potential for creativity, learning, and purpose. The technology itself won’t decide which path we take. We will. I explore this idea in my latest piece: “Will The Rise Of AI Mean The ‘Termination’ Of Humankind?”

#AI #AIAgents #FutureOfWork #Startups #Technology #Innovation#Society

Most AI discussions focus on models. But in healthcare, impact comes from systems. The Clinical Agent Stack reframes AI as a coordinated, multi-layer system where humans and machines work together.

Trust in the digital economy stems from reputation, institutions, incentives, and culture, not just cybersecurity or technology layers

Moving beyond rigid rituals to true adaptive planning. A reflection on what Agile actually teaches us about navigating complexity, uncertainty, and continuous delivery.

Scenario planning helps businesses navigate uncertainty by preparing for multiple outcomes, shifting focus from prediction to resilience

Geopolitical energy shocks raise oil prices, tighten capital, and reshape startup innovation toward efficiency, resilience, and AI energy challenges

An examination of how digital platforms, AI, and automation have reshaped economic structures — often reinforcing the divides they promised to close. A critical look at tech's unintended consequences.