When people ask, “So, what’s your idea?”, they usually mean: what’s the clever thing you’re building?

But the longer I’m in this, the more I see that good companies don’t exist because of “ideas.” They exist because they sit on a very specific, very real problem and solve it better than anything else. They grow only when that problem is shared by more people, or when they move closer to something the world must get right.

I’ve heard this since college. “Solve a painful problem.” I thought I understood it. In reality, most of what I’ve done so far is: start from an idea I like, then reverse-engineer a “problem” that makes the idea look justified.

Sometimes you’re not solving a problem; you’re imagining one. Or you discover a small, local pain and prematurely decide, “This is it.” Then you stop looking.

As I’ve run more discovery calls, that illusion has started to break.

Take manufacturers.

The first time I landed on “AI SDR for Chinese factories” was at a Chinese New Year dinner in February 2025. My uncle runs Meta ads for his sanitaryware factory. His complaint was clean and sharp:

He spends on ads.

He gets a lot of inbound leads.

His team can’t handle the volume, so he outsources follow-up for $7,000 to a cold calling team (that’s pretty expensive and he definitely overpayid for it, especially for a job like this in China)

So, on paper, that’s a standard “painful problem.” In practice, he just landed at a worse problem: after paying $7k the outsourced team couldn’t actually talk to his customers.

They had broken English, no Spanish, no Portuguese; meanwhile his campaigns targeted South America. So the calls basically started with “Hello, how are you?” and collapsed immediately. Leads hung up, or stumbled through English both sides barely understood.

I gave him something that clearly made sense: a multilingual AI sales rep to qualify and follow up. We hacked it together with Vapi, shipped it, and his conversion rate went up.

I took that as validation. I got attached. Suddenly, I wasn’t exploring a space; I was “the guy building AI SDRs for factories.”

I told myself I was “validating a problem.” In reality, I was looking for more people who fit my uncle’s template so I could validate the product I already liked. That’s how you get stuck in a local minimum.

Because yes, almost every factory would like “more sales.” But that doesn’t mean they all have my uncle’s problem.

If you talk to 10,000 factory owners and only 5 of them run Meta ads and are actively looking for a better way to qualify those leads, then you’ve optimized your life around a tiny niche without realizing it. You’ll be “busy building” and “getting traction,” but maybe on something structurally small, or worse, misaligned with how the rest of the market actually works.

As I kept talking to more suppliers, their reality turned out to be more diverse:

• “We’ve always relied on word-of-mouth. Now an old customer switched to another supplier with better price and quality.”
• “We depend entirely on Alibaba. If we stop paying for exposure, our pipeline dies.”
• “We have tons of old leads, but nobody has time to follow up properly.”

At the same time, you hear: “Alibaba isn’t working anymore. We spend a lot on ads and the lead quality is bad.”

Those sound similar on the surface (“sales not growing”), but they point to different underlying problems:

• Channel concentration risk vs.
• Unit economics of paid traffic vs.
• Lack of sales process vs.
• Lack of talent vs.
• Product and pricing competitiveness.

If I lazily label all of this as “factories want 拓客 (lead gen),” I end up with a fake problem statement and a generic solution:

“We’re building an AI sales rep that helps you do outbound and 拓客 24/7.” (sorry, but what the heck is outbound?)

It’s just uselessly broad and doesn’t force me to answer:

• Which factories?
• In which export corridors?
• With what existing channels?
• With what willingness to change behavior and share upside?

After talking to more than ten factory owners, some patterns became clearer: they say their Alibaba and tradeshow ROI is dropping, but the underlying issue is: They are completely reliant on Alibaba and tradeshows and openly say they want to diversify.

Almost nobody phrased their core issue as “I need an AI SDR.” That’s my language, not theirs.

Steve Jobs responding to a guy that insulted him because Steve didn’t prioritize the tech; Jobs handled it very well and gave a masterclass on why we start with the customers and be obsessed with the problems, not the tech/product.

From that, a more honest problem statement starts to emerge:

“There is a group of export-oriented factories that are over-dependent on Alibaba and fairs. Their CAC is rising, lead quality is inconsistent, and they don’t have the internal talent to build alternative channels.”

Now my pitch can shift from generic AI magic to something more grounded:

“We help you reduce your dependence on Alibaba while still getting qualified leads - and you don’t pay us until an order is closed.”

That’s still not the final form. It’s just noticeably sharper than:

“We’re building an AI salesperson that proactively 拓客 for you 24/7.”

Because “拓客” is not a problem; it’s more like a bucket. “Why is my sales not growing?” is not a problem; it’s a question that hides ten different problems.

The humbling part is realizing this:

“Solving a painful problem” is too broad to be useful.

You only really start solving a problem when you can describe it so concretely that:

A specific type of person hears it and says, “That’s me.”

You can tie it to a budget line and a metric.

You can point to existing, unsatisfying attempts to fix it.

I’m still early in this process. The insights I have are shallow compared to what they need to become. But at least the delusion is breaking: I’m not “the guy with the AI SDR idea” anymore. I’m back to being a student of what’s actually broken in how factories find and grow their business.