I Went to Alibaba and Came Back with Three GenAI Use Cases for Cross-Border E-Commerce

A while ago, I flew out to Hangzhou specifically to visit Alibaba's new campus in Xixi.

Visiting tech companies is probably my favorite hobby. This time, my mission was clear: Alibaba's international e-commerce group, AIDC.

Lazada, Daraz, Trendyol, and AliExpress — which had been riding high the past couple of years — all fall under its umbrella.

Why was I interested in this business? Because Alibaba's international arm has made a lot of waves over the last two years. In May 2024, AIDC's revenue for a single quarter grew 45% year over year. Who was the biggest contributor? AliExpress Choice.

There was also a management reshuffle. That same November, Jiang Fan — who had been running AIDC — also took charge of the domestic e-commerce business. All of Alibaba's commerce was now under his sole command.

I'd looked at the numbers and met plenty of people. But what I really wanted to get to the bottom of was something else.

What Does GenAI Really Mean for Cross-Border E-Commerce?

Put simply, it means putting large language models to work on the tedious, thankless job of selling goods overseas.

Cross-border is a huge market, and it's growing fast. What I especially wanted to know was just how much faster AI could make it.

One detail worth flagging right away: Alibaba has set up a business unit dedicated solely to GenAI. It doesn't sit inside group headquarters, and it doesn't sit inside Alibaba Cloud — it stands entirely on its own.

After the full tour, I came back with three genuinely useful applications. Let me walk through them one by one.

Use Case One: Start by Making Merchants More Efficient

Alibaba runs a marketplace. There are two kinds of people on it: the merchants and brands on one side, and the buyers on the other. Among the buyers are consumers, but also businesses. And don't forget — Alibaba.com runs a whole separate cross-border B2B line of business.

So where does AI get applied first? On "its own people": making merchants and brands more efficient.

Why start there? For safety. Something used internally, even if it messes up, can't hurt the customers. And efficiency is exactly the kind of job AI is naturally good at.

But cross-border is unlike China's domestic market — it piles on a heap of problems that don't exist at home. The biggest one: language.

AIDC is one network spanning more than 200 markets around the world. More than 200 markets means more than 200 languages. You need language to take orders, and you need language for customer service. This isn't a small problem — it's a huge one.

The simplest approach is to use AI to translate product listings and manuals into every one of those languages. It works, but it's nothing out of the ordinary.

What really catches the eye is cross-border marketing.

Sales copy gets translated into the local language, product images get redone in different versions, and you can even roll out marketing videos in dozens of country-specific versions.

But the real skill here isn't translation. An image for the Chinese market and an image for the Kenyan market follow completely different playbooks. Different buyers, different habits, different tastes.

So when GenAI is brought in for cross-border marketing, the point is to push the conversion rate as high as possible in every single market. Saving a few plane tickets for photo shoots is just a bonus.

The most striking part of all this is video.

Think about it: a Korean skincare brand needs marketing and educational videos for more than 50 countries around the world — different languages, different aesthetics, and they still have to actually work. How do you pull that off?

You call in AI digital humans.

Where on earth would you find 50 real presenters for 50 countries? AI can create them. For Kenya, you build an African presenter who introduces the product in the local language, covers the current promotion, and picks up the hot topics of the past few days.

Doesn't that get you buzzing, just to think about it?

Use Case Two: An AI Advisor to Take Merchants Overseas

The second application is every bit as interesting.

A lot of merchants are already successful at home. But reach a certain point and you naturally look up and hit the ceiling: so what's next?

Going overseas. Everyone can say that easily.

Yet ask for specifics and they get stuck on the details: which country? Which customer base? Why win on someone else's home turf?

Most small and medium enterprises can't answer those three questions.

What's missing? Someone who heads up international business. I'm often dragged in to play that role myself. And it's data that's even scarcer. If you don't even know which segment to target, how can you talk about going overseas?

AIDC has shipped a chatbot — an AI advisor for going overseas. Tell it which market you want to enter, what you want to sell, and how you plan to go about it, and it will analyze and advise. Why should you trust it? Because behind it sits the transaction data of Alibaba's international e-commerce. When it comes to what people around the world are buying, Alibaba is probably one of a very few who know best.

Once you've talked it through, the natural next step is Alibaba.com.

Alibaba.com is a born-and-bred cross-border B2B platform that matches small and medium enterprises around the globe with other small and medium enterprises. A small factory in Texas orders parts from a small factory in China; that Chinese factory may, in turn, buy raw materials from the Philippines. It's a chain that just keeps turning, link by link.

But supporting that chain is the hard part: buyer-seller credibility, factory and capital verification, escrow, global logistics, customs and duties, compliance filings across dozens or even a hundred countries, and endless disputes that are hard to sort out.

Alibaba.com has long been able to manage this whole package. Pair it now with an AI advisor, and it's like giving a veteran driver a navigation system. It's a combination I genuinely have high hopes for.

Use Case Three: The Underrated Real-Time Translation

The third use case isn't flashy, but it carries the most weight.

The language problem mentioned earlier is now being tackled by real-time translation. Domestic software like DingTalk already handles Chinese-English translation quite well. But large models push it to a new level: multimodal and real-time.

What does "multimodal" mean? It takes in text, images, video, and audio alike.

Alibaba has a tool called Qwen Tingwu that's very good at real-time multimodal translation — Chinese in, English out, all in real time. At this pace, "carrying a virtual translator around in your pocket" really isn't far away.

So what does cross-border use it for? Externally, it's sales and customer service, which we've covered. The real easter egg is internal — almost every country needs a few people to handle local customer service and operations. The more countries you add, the messier internal communication gets. What language should a German coworker speak with a Mexican coworker in a meeting? With real-time translation, the moment this side finishes speaking, the caption appears on the other side. The efficiency stuff is real.

What's more, translation is now starting to understand what "your line of business is talking about" — no longer word-for-word translating.

Conclusion

So those are the three use cases I brought back from the Xixi campus. To put it in one sentence: AI won't go overseas for you, but it can lower the cost of going overseas and speed it up.

For many companies, the cross-border path is genuinely worth taking.

This trip to Xixi wasn't a wasted trip. Let's call it a day here.