You Vibin'?

Fun fact: I’ve been a personal finance nerd since 2017, diligently tracking portfolios, budgets, and various "whatnots" with the discipline of a monk.

Not-so-fun fact: I’ve been doing all of this via a trusty spreadsheet for that entire time.

Now, don’t get me wrong. That file has seen as many iterations as a Product Manager has backlog items. And while I’m fully aware I could "digitally transform" my life with any number of sleek, modern apps, I’m a total control freak about my data. I’ll take my manual entries and data privacy over a third-party integration any day. Call me old-fashioned; I’ll be over here adjusting my formulas.

Last night, though, I was in the mood for an experiment. It’s the start of the year, and I was looking for ways to level up my process. A persistent pain point has been managing cash flow, accruals, and the inevitable repayment lag across multiple credit cards. It dawned on me: I could solve this by building my own sleek app via vibe coding. I’d been meaning to take Lovable and Base44 for a spin anyway, so I dove in.

One word. Actually, no, one emoji: 🤯.

I had two functioning, full-stack web apps in less than ten minutes. Two additional features were live in less than half an hour.

As a product person, I have some initial thoughts on this experiment (operating word being "single," since more practice yields different results):

1. I was the user, and I knew my desired journey. I ensured the problem was clearly captured in the prompt. I wonder how these platforms would’ve proceeded if the initial text was vague. Would they have pushed back? Would they have asked what problem I was solving and why it mattered?

2. Both Lovable and Base44 platforms suggested functionality. However, I found that most of them were superfluous. An experienced product person would have anticipated context-based needs.

3. To be fair, both platforms made design and implementation decisions for the features I did ask for. I only broadly specified "file upload" and "break identification," yet both built out logical workflows for data mapping and visualization without being handheld.  

   
   

In the end? I used neither. Even with airtight security settings, I realized I’d still be uploading my raw data to their platforms, and the control freak in me just wasn't ready to let go. But it’s still a happy ending: I found a middle ground by vibe coding via Gemini to build a localized, fit-for-purpose, fit-for-need solution. Et voilà. 

It’s been about a year since Andrej Karpathy introduced "vibe coding" into our collective awareness, and it’s clearly moved past being just a viral tweet. In fact, a huge percentage of startup founders are already seeing the vast majority of their codebases being AI-generated according to a terrific Y Combinator podcast episode I saw just today. I highly recommend spending 31 minutes and 33 seconds on that video as it covers fascinating insights on solving product problems vs. system problems, how the word "technical" now may shift to meaning a background in math and physics and not necessarily classical computer science training, the continuing importance of hardcore software engineering especially after achieving product market fit, among many others.

The key takeaway: We’re all product people now. We are creating value by intent, not necessarily by syntax.

What an age to be alive in. Or should I say, what an age to be vibin'. 🤘