What Was Your First Program Run?

Reading through the breakdown on modernizing four-decade-old scientific code with Gemini brought back a flood of memories from where my own journey in computing began.

My very first program wasn’t typed into a modern IDE or executed in a cloud container. It was coded in Fortran on a physical deck of punchcards.

If you weren’t there, it’s hard to capture the sensory experience of computing back then:

– The rhythmic, metallic clatter of the card punch machine under fluorescent lights.
– The meticulous care taken not to drop the deck – because a dropped stack meant hours of re-sorting cards.
– The quiet anxiety pacing outside the glass-walled computer room, handing over your batch job and waiting to see if it compiled.
– The tangible satisfaction (or sinking feeling of a syntax error on card 42) when unwrapping that wide, continuous blue-and-white striped tractor-feed printout.

Fast-forward to today: AI models like Gemini 4 Argon are actively ingesting, deciphering, and modernizing complex, 40-year-old scientific codebases in seconds – translating decades of legacy domain logic into modern, optimized, and maintainable architectures.

What once took a long turnaround time, physical logistics, and manual tracing across paper printouts is now resolved within an interactive inference window.

The fundamentals of problem-solving haven’t changed, but the leverage we have as builders has shifted beyond anything we could have imagined standing outside that computer room.

Technology advances at breathtaking speed, but it’s always grounding to look back at the paper and punchcards that paved the path forward.

For those who lived through the batch-processing era: what was your very first tech stack or program run?

Don’t Follow Your Passion…

“Don’t follow your passion… follow your effort.”

That’s one piece of advice I find myself sharing often with startup founders and student mentees.

It is easy to tell the world – and yourself – that you’re passionate about something. Passion sounds inspiring in a pitch deck, during coffee chats, or on an “About” page. But look closely at your calendar and your task list: are you actually putting in the relentless, unglamorous work to pursue it?

The ideal alignment, of course, is pouring massive effort into what naturally excites you.

When that’s not happening, flip the question:

Look at where your hours, focus, and energy actually went over the past month. If you are consistently drawn to solving a specific problem, mastering a tricky skill, or grinding through a complex project – even when no one is asking you to – then pay attention.

Your effort doesn’t lie. It reveals your true curiosity long before your resume catches up.

Often, passion isn’t the starting line; it’s the byproduct of competency built through effort.

Look back at your calendar: where did the bulk of your real effort go last week?

Your AI Life Story

I have a slightly unusual habit whenever I travel abroad for work: I always pick up a local newspaper and turn straight to the obituaries.

It’s fascinating how much this single section reveals about cultural differences.

In Hong Kong, an obituary is typically formal and concise – often dominated by an extensive list of surviving family members and lineages. In many Western papers, the focus is deeply human: quirky anecdotes, formative struggles, and life lessons. It captures a life in full color.

Lately, this habit made me wonder: What will an AI-generated “Life Stories” look like for our generation?

If an AI agent synthesizes our digital footprint – our emails, commit histories, published papers, and social feeds – will it summarize us by our titles, pedigrees, and corporate milestones? Or will it actually uncover the quiet, defining moments that made us who we are: the projects that failed, the mentorship given off the record, the late-night problem-solving?

Technology often defaults to cataloging achievements. But what makes human history worth reading isn’t the clean outcome; it’s the friction along the way.

As we build tools that summarize and automate our world, we should remember to measure what truly matters. Celebrate the messy, authentic journey – not just the curated title.

What Are Useful AI Use Cases?

What are your AI use cases?

* Drafting emails?
* Summarizing 50-page industry reports?
* Generating social media content?
* Prototyping a quick game or building personal workflow tools?

Most of our day-to-day conversations around AI naturally center on personal productivity and speed – getting through our inbox faster, sharpening our code, or automating routine tasks. And that alone has transformed how many of us work.

But then you see projects that remind you of what AI can do on a planetary scale.

A fantastic example is Cathay Pacific’s collaboration with Google and Contrails.org (shoutout to Oliver Haas and the team for sharing this journey).

By applying AI models to predict where contrails – the heat-trapping cloud streaks left behind by flights – are likely to form, flight operations teams can make minor altitude adjustments to bypass them. It’s an elegant, data-driven approach to tackling real-world aviation climate impact in real-time.

Efficiency gains are great. But using AI to make our planet a better, more sustainable place to live? That is where the real inspiration lies.

Curious to hear from you all : Beyond everyday productivity and automation, what is the most impactful or unexpected real-world application of AI you’ve seen lately?

Play The Right Game

Success is no longer determined by getting better at playing yesterday’s game using AI. It’s defined by whether you’re playing the right game.

Almost everyone now has access to powerful LLMs, GenAI tools, and intelligent workflows. As a result, content is abundant: perfectly structured emails, solid analytical reports, polished slide decks, and slick video assets.

Yet, when output becomes effortless, distinctiveness disappears. Without clear human intent, much of what we see feels generic, pale, and transactional.

AI amplifies capability, but true differentiation comes from how we apply it.

I spent this past week for an intensive training series. We moved beyond simple prompting and coding, to focus on what matters most for scalable, enterprise-grade AI:

+ Output Quality & Authenticity: Bringing critical thinking and domain expertise back into AI-generated deliverables.

+ Agentic System Design: Moving from standalone prompts to multi-agent architectures.

+ Evaluation & Security: Ensuring governance, risk mitigation, and continuous quality measurement.

The tech landscape is evolving rapidly. The goal isn’t just to produce more, but to figure out what to do – it’s executing with quality, judgment, and human nuance.