Vibe Coding
Posted: February 13, 2025 Filed under: Development, Trick | Tags: paradigm Comments Off on Vibe CodingTwo weeks ago I started to noticed about something that’s both fascinating and slightly unsettling: vibe coding. And no, it’s not about writing code while listening to lo-fi beats (though that sounds nice too).
Vibe coding, as coined by Andrej Karpathy, is essentially “fully giving in to the vibes” and forgetting that code even exists. Instead of writing lines of code, you’re having a conversation with AI about what you want to build. It’s less about “how” and more about “what”.
Think about it: instead of googling Stack Overflow (that was so 2022, who’s still do that?) for the millionth time about that one CSS property you can never remember, you just tell an AI, “Make this button look cooler.” And it… just does it?
Even wild, you can make it independently execute 10 steps by itself.
The wild part is that it works. People are building entire applications without writing a single line of traditional code. It’s like having a really smart dev friend who does all the technical heavy lifting while you focus on the creative vision. I been using voice/STT as interface (because typing is slow) and AI IDE like Cursor since last year (even longer with voice as interface – almost 10 years till this date, can you believe it? **Dragon Dictate, hello!), and I agreed it was new paradigm even though I felt like I was cheating.
But here’s what fascinates me most: we’re moving from a world where coding was about speaking the computer’s language to one where computers are learning to speak ours. It’s not about syntax anymore—it’s about vibes.
Is this the future of software development? Maybe. Will it replace traditional coding? Probably not entirely. But it’s definitely changing how we think about building software. – I am firm believer the future is also agentic.
For now, I’m both excited and slightly terrified to see where this goes. After all, sometimes the best vibes come from understanding what’s happening under the hood.
What do you think about this shift? Are you ready to give in to the vibes?
ROI Over Use Cases
Posted: February 13, 2025 Filed under: Product | Tags: perfectionism, pm Comments Off on ROI Over Use CasesI’ve often found myself in what I now call “researcher’s syndrome” – that relentless urge to cover all angles, explore every possibility, and chase down every use case. As a PM, it felt natural: the more I knew, the better decisions I could make. Or so I thought. This is the story of how I learned that sometimes, less truly is more.
The Realization
In my early PM days, I was that person with sticky notes everywhere and a roadmap that looked like a tangled web of possibilities. I believed success meant creating products that solved every conceivable user need. My research was exhaustive, my feature sets were bloated, and honestly? My team was exhausted.
It wasn’t until a mentor @chuazm bluntly told me I was “trying to boil the ocean” that something clicked. The truth hit hard: in trying to solve everything, I was solving nothing particularly well. I also pickup new approach to first principal thinking with our EM @lsshawn.
Why Less Is More
Here’s what I’ve learned about focusing on what truly matters:
- Clarity Over Clutter
When we focus on fewer features, we gain clarity about what our product actually stands for. It’s not about doing less work – it’s about doing better work. Apple’s approach to product design has been my north star here. Their relentless pursuit of simplicity isn’t just aesthetic; it’s strategic. - Depth Over Breadth
Think about Slack. They could’ve tried competing with full project management tools or email clients, but instead, they focused on one thing: making team communication seamless. This laser focus allowed them to excel where it mattered most. - Strategic Alignment with ROI
The shift from “What else can we add?” to “What will deliver the highest impact?” transformed my decision-making process. Not all features are created equal – some might delight users but barely move the needle on business metrics. Learning to differentiate between “nice to have” and “need to have” became crucial.
The Philosophy in Practice
Here’s how I’ve integrated this mindset into my work:
- Start with Why: Every feature needs a clear purpose tied to user needs or business outcomes.
- Kill Your Darlings: Regular roadmap audits are essential. Be ruthless about removing features that don’t add significant value.
- Focus on Outcomes: Shift your thinking from features shipped to problems solved.
One of my proudest moments was when our team chose to improve an existing workflow used by 80% of our users instead of building a flashy new feature. The result? A 30% increase in user satisfaction. Sometimes the best solutions aren’t the most exciting ones.
A Closing Thought
Minimalism in product management isn’t about taking shortcuts – it’s about making deliberate choices that amplify impact. It’s recognizing that our resources are finite and choosing to use them wisely.
If you’re feeling overwhelmed by possibilities or paralyzed by options, remember this: success isn’t measured by how many features you ship, but by how well you solve real problems. Sometimes, doing less is the bravest – and smartest – choice we can make.
I’ve learned that my initial instinct to cover every base wasn’t just about thoroughness – it was about validation. But true impact comes from focusing on what matters most and executing it exceptionally well.
What could you achieve by doing less, but doing it better?
P/S: I got inspiration to write this after watch Arseny Shatokhin video on AI agent development.
To Code or Not to Code, Revisited
Posted: February 9, 2025 Filed under: Uncategorized Comments Off on To Code or Not to Code, RevisitedSo, I ranted about code in static papers a while back (someone asked me about this – so I’d figure allow me to do a follow up), but what if your code is the real star of the show? Can you still sneak some in without drowning readers in curly braces?
Yes.
If your snippets are absolutely critical—like the secret sauce that everyone needs to see—fine, slip in a few lines. Just make them short and sweet. No one wants to navigate a labyrinth of semicolons; your readers only care about that one essential function or loop that proves your point. But don’t stop there: you still need to explain why those lines matter. If you’re showing off a special parsing trick or a brand-new algorithm, tell us why it’s special—don’t just paste it and pray.
Most importantly, provide a link to the full code in its natural habitat, whether that’s GitHub or another platform. Readers who truly want to dig deeper will thank you for letting them see—and run—the real thing, rather than forcing them to squint at a crammed PDF. Give context, give clarity, and keep the clutter to a minimum. After all, no one’s going to appreciate a code dump… not even your future self re-reading that paper.
Keep it stupid simple. KISS.