Let me ask you something. If you run a wholesale business, does every single step—purchasing, inventory, accounting, chasing payments—feel like it’s slowly draining your energy? You hire people, costs go up. You don’t hire, you kill yourself. At some point, the business gets stuck.
Most people assume AI has nothing to do with this kind of grind. But Zhang Feng proved otherwise in just three months.
He started from a 5-square-meter women’s clothing stall in Beijing’s Zoo Market, worked for nearly a decade, now serves 15,000 retail stores, and hit 40 million yuan in revenue last year. Today, 90% of his energy goes into AI. Not because he’s chasing trends—because he’s been burned by every single pain point, one by one, and found a way to patch them with tech.
How?
Step 1: Start with the worst pain point—even an extra two hours of sleep is a win.
Most people jump straight into building big models or smart agents and get nowhere.
Zhang’s entry point was tiny: his wife’s daily ordeal. Every time she went to purchase, she had to manually enter 300–400 SKUs into the system. Two hours of repetitive data entry, between different systems, error-prone, soul-crushing.
He used OCR to scan the invoices and auto-fill the ERP. Two hours became 20 minutes.
His wife got two extra hours of sleep every day. For a wholesale clothing boss, that’s gold.
Step 2: Turn your internal tool into a product—let your customers validate it.
Problem solved for himself. Then he noticed something: his 15,000 downstream retail store owners were stuck in the same trap—shooting videos, drafting scripts, editing content. Each step was a time sink, each had a learning curve.
He packaged the AI content tool he’d built for himself into a product called “Ji He Partner.”
Core features: one-click templates, auto-script breakdown, generate model photos from a flat lay image. Used to take five steps to upload clothing; now one photo does it all.
Within a month, accounts using this framework hit three million-view videos. One reached 20 million views, gained 30,000 followers.
Compare that to his wife’s own account, which grew only 1,000–2,000 followers per month through manual effort.
That’s the best control group you could ask for.
Step 3: Follow customer feedback, and let it lead you to a new revenue stream.
Once the product launched, users came with problems.
The earliest image-generation feature required uploading front, back, top, bottom separately, choosing a model, pose, scene. Too many steps. Zhang cut it down: upload one flat lay, pick a model, let AI handle the rest.
While refining, he hit another realization: offline sales data was completely invisible. Senior salespeople closed deals, new ones couldn’t learn—no one knew where the gap was.
He went to Huaqiangbei (Shenzhen’s electronics hub) and found a smart badge with built-in recording, auto-slicing, cloud sync. It tracks each salesperson’s performance and the full conversation flow.
899 yuan each. He ordered 100 units for a test. That’s the third product line, almost by accident.
The bottom line: Traditional business owners don’t need to know code to make AI work for them. They just need to feed AI with the pain points they know best—then take one step at a time.
Start with the one thing that wastes your time every single day. Solve it. Then listen to your customers. The next curve will grow by itself.