People looking for electronic parts online tend to arrive in the same state. They know the part number but not where to buy twenty of them. They are buying for a company, so they need a tax invoice. They want a quote by email and the site tells them to phone instead. One of our subsidiaries, Dukwoo Trading, imports and sells electronic parts, so we get these questions every day. We answered them by building the store, and the part worth writing down is how much of the building was done by AI.
Three things worth checking on any parts store
Whichever store you land on, three checks settle it. Does it search by part number, because parts are bought by number rather than by name. Can you get a quote for a small lot, twenty pieces rather than a full reel. Does it issue a tax invoice, because for a company purchase that one is pass or fail. We built the mall against those three. You put several part numbers into one basket, send the quote request, and the answer comes back by email.
Nobody was going to write 662 pages by hand
The obstacle was the page count. Every item needs its own page to be findable at all, and writing them one at a time is a job measured in months. So the work of turning item data into pages went to AI, and my part was inspecting what came out.
| What was produced | How many | How long it took |
|---|---|---|
| Item pages | 662 | One day |
| Components among them | 117 | Same pass |
| Industrial supplies among them | 545 | Same pass |
| Descriptions rewritten later | 122 | One day |
Ranked third, and nobody clicked
Once the site was live I went and read the search data, and something stood out. Certain part numbers were sitting at position three with zero clicks. Position three means people are seeing it. Zero clicks means the description underneath is not doing its job. So I picked those part numbers out and had the descriptions rewritten: 122 item descriptions fixed in a day, AI drafting and me checking that nothing was wrong. Done by hand, that is a job you never even start.
What the AI did, and what people did
The split came out cleaner than I expected:
- AI builds the item pages and writes the descriptions.
- AI reads the search data and drafts the replies that go back out.
- People decide which items to carry and how they are priced.
- People decide which customer questions the site should answer, and check the output before it goes live.
If you know the part number, just ask
Putting AI into a company is not a grand project. It is turning months of repetition into a day and leaving people with the judgement, which is also how I decide which of my people get which tools in the first place. This one store is my evidence. If you know the part number, you can put it in the basket at the Dukwoo mall and send a quote request, and the answer comes back by email.
Months of repetition became a day. What was left for people was the judgement.
That is the whole story of this build, and I suspect most companies are sitting on the same opportunity without naming it. So here is the question I would ask in your shoes. What is the repetition that someone in your company has been doing by hand for months?
