← Back to Blog

From Vietnam Hotel to Production: How Customer Discovery Actually Works

Real customer discovery doesn't happen from your desk. Learn how a hotel manager in Vietnam revealed the brutal truth about finding problems worth solving.

I'm sitting in a beachside hotel in Hoi An, Vietnam, supposedly on a business trip for the Dubai-based VC firm I consult for. It's 6 PM, and I'm watching the hotel manager hunched over his laptop, squinting at a stack of paper invoices. Click, type, click, type. He's manually entering each line item from 75 customer bills into Excel.

This is how I discovered my next product wasn't going to come from ChatGPT, Twitter, or a ProductHunt trending page. It was going to come from watching a guy waste two hours of his life every single day doing something a Python script could handle in thirty seconds.

The Problem with Building from Your Desk

Let's be honest about how most of us build products:

  1. Browse Twitter, see someone's MRR screenshot
  2. Ask ChatGPT for "SaaS ideas that could make $10k/month"
  3. Check ProductHunt to see what's trending
  4. Build a clone with "AI" slapped on it
  5. Wonder why nobody's buying

I know because I did exactly this with my first five projects. I built what I thought people needed, what online communities said was "hot," what AI suggested would be profitable.

Zero sales. Zero retention. Zero product-market fit.

The brutal truth? You can't discover real problems from your desk chair. ChatGPT doesn't know what a boutique hotel manager in Vietnam struggles with. YouTube tutorials won't tell you about the accountant still using printed ledgers. Hacker News won't surface the thousand small businesses drowning in manual processes.

The Hotel Manager Who Changed My Perspective

I wasn't in Hoi An to build products. I was there evaluating potential investments for our portfolio. But when you're a developer, you can't unsee inefficiency.

Every evening at 6 PM, this manager would sit at the small desk behind reception with:

He'd type every single item into Excel. Guest name, room number, service, amount, tax, date. Row by row. Cell by cell.

"How long does this take you?" I asked.

"Two hours, sometimes more if busy day," he replied in broken English. "But must do. Boss wants daily report."

Two hours. Every. Single. Day.

That's 730 hours a year. Almost a full month of his life spent copying numbers from paper to screen.

The 30-Minute Solution Nobody Was Selling

Here's what killed me: The solution was stupidly simple.

I spent 30 minutes that evening writing a Python script using:

The next day, I showed him how to:

  1. Take photos of the invoices with his phone
  2. Drop them in a folder
  3. Run the script
  4. Get a perfectly formatted Excel file

His 75 invoices? Processed in under 2 minutes.

The look on his face wasn't just gratitude – it was disbelief. "Why nobody sell this?" he asked.

Good question. Why wasn't anyone selling this?

Why Real Problems Hide from Developers

Because real problems don't announce themselves on social media. They don't trend on ProductHunt. They don't appear in "Top 10 SaaS Ideas for 2025" lists.

Real problems are:

The hotel manager didn't wake up thinking, "I need an AI-powered invoice automation tool." He woke up thinking, "I have to do that Excel thing again tonight."

He wasn't searching for solutions because he didn't know they were possible. He wasn't complaining on Reddit because he doesn't use Reddit. He wasn't filling out your customer survey because you never met him.

The Myth of Desk-Based Discovery

We've been sold this lie that customer discovery means:

But here's what actually works:

What I Learned from Stepping Outside

Since that trip to Vietnam, I've made it a rule: No new product ideas from my desk. Every successful tool in my portfolio came from being somewhere I wasn't supposed to be building software:

SHRP (PDF data extraction): Born from that hotel in Vietnam, now serving 50-100 businesses daily who have the same problem.

MedGPT (medical translator): Discovered while getting a health checkup in Bangkok, watching doctors struggle to explain reports to foreign patients.

iKrypt (secure contracts): Came from sitting in a law office in Dubai, watching them email unencrypted contracts back and forth.

None of these ideas would have come from ChatGPT prompts or Twitter threads.

The Cost of Comfortable Research

Every minute you spend asking ChatGPT for business ideas is a minute you're not discovering actual problems. Every ProductHunt launch you analyze is a solution that already exists. Every YouTube tutorial you watch is teaching you to build what everyone else is building.

Meanwhile:

These aren't sexy problems. They won't get you on TechCrunch. But they're real, painful, and worth paying to solve.

How to Actually Find Problems Worth Solving

1. Travel (or just leave your neighborhood)

2. Work temporary jobs

3. Shadow professionals

4. Look for paper

5. Find the 6 PM workers

The Uncomfortable Truth About Building Products

If you're a young developer building web apps because Twitter told you it's the path to freedom – you're playing a game you can't win. You're competing with thousands of others building the same AI wrappers, the same productivity tools, the same solutions to problems that might not exist.

The developers making real money? They're solving unglamorous problems for people who don't use Twitter. They're automating workflows for businesses that still use Windows XP. They're building boring tools that save real time for real people.

But to find these opportunities, you need something that no amount of coding skill can replace: life experience.

You need to have worked different jobs. Traveled to different places. Met people outside the tech bubble. Seen how the world actually operates, not how Silicon Valley thinks it should.

Stop Building from Assumptions

That hotel manager in Vietnam now saves 2 hours every day. His boss is happy. The data is more accurate. And he told his hotel manager friends about it.

That's 5 more hotels now using my tool. Not because I had great marketing. Not because I went viral on ProductHunt. But because I solved a real problem that actually existed.

Your next successful product isn't hiding in ChatGPT's suggestions. It's not trending on Hacker News. It's not in that YouTube tutorial about "10 SaaS Ideas That Print Money."

It's in a hotel in Vietnam. A clinic in Bangkok. A law office in Dubai. A restaurant in your own city.

But you'll never find it from your desk.

The Challenge

Here's my challenge to you:

This week, spend 2 hours somewhere you don't belong. A local business. A government office. A medical clinic. A law firm. Don't go to sell anything. Don't even mention you're a developer.

Just watch.

Watch for:

Then build something small. Something specific. Something that solves one person's exact problem.

Don't worry if it's not "scalable." Don't worry if it's not "disruptive." Don't worry if VCs wouldn't fund it.

The hotel manager in Vietnam doesn't care about your TAM or your AI moat. He cares that he gets to go home at 6 PM instead of 8 PM.

And he'll happily pay $50/month for that privilege.


Currently traveling through Southeast Asia, finding problems worth solving. If you're stuck building solutions nobody wants, maybe it's time to close your laptop and open your door. The real world is full of inefficiencies – you just have to be there to see them.

P.S. - That hotel in Hoi An? They're now my most enthusiastic customer. Turns out solving real problems creates real advocates. Who would've thought?