Semarlin Jianxia: How AI Is Transforming the Way Thousands Work

Lin Jianxia is a post-80s professional—slim, sharp, and with a strong, confident voice. When she talks about DingTalk AI tools, she lights up like she’s catching up with an old friend, sharing every detail with enthusiasm. Her colleagues affectionately call her “Sister Xia.” In their eyes, Sister Xia isn’t a tech expert—but she makes AI feel far from mysterious or out of reach.

But she’s not the stereotypical tech guru. Just three years ago, she was a business employee at Semir, having spent more than a decade in roles like sales, planning, and merchandise management—fields that had little to do with technology. No one could have predicted that today she’d become the company’s “AI implementation expert,” bringing a groundbreaking tool into the daily workflows of thousands of coworkers.

Lin Jianxia has an unusually strong hands-on passion for AI. She says it’s not that she’s particularly special—it’s that opportunity found her, and she’s happy to take on challenges beyond her KPIs.

This “selfless” drive stems from a deep conviction within her: she firmly believes AI is the path to the future. She’s deeply impressed by the concept of “collective imagination” from Sapiens: A Brief History of Humankind—how humans use language to tell stories, creating shared beliefs that form the foundation of civilization. She sees AI as creating a new kind of “collective imagination.” “The DingTalk AI tables and AI assistants we use today don’t require years of learning code just to write ‘Hello World.’ They’re tools that anyone can pick up and use right away.”

Her extensive cross-functional experience has given her a deep understanding of the limitations of traditional workflows. But without a technical background, how did she seize the opportunity to master these new tools and transform work processes? And how did she bring AI to her colleagues, bridging the information gap?

Here’s Lin Jianxia’s story—

Three years ago, I was transferred to Semir’s Digital Center—and by chance, I ended up taking responsibility for rolling out AI across the entire company. I don’t come from a technical background; I relied entirely on my sensitivity to digitalization and my interest in AI to figure things out step by step.

The moment I truly realized the power of AI came in early 2023, when ChatGPT 3.5 exploded onto the scene. At the time, my daughter’s school was planning a holiday video, and as a parent committee member, I needed to write lines for 25 parents in the class. It was a tedious task—researching, piecing together sentences, and crafting the script. On a whim, I borrowed a colleague’s phone and asked ChatGPT for help. To my surprise, the content I needed appeared in just a few seconds.

I was blown away—AI felt like magic. I began wondering: In the future, will AI become as universal and accessible as language itself?

In my years at Semir, I’ve worked in sales, planning, and merchandise management, so I know all too well the pain points of traditional work methods. That’s when I started asking myself: Can AI help us solve these problems?

The first time I applied AI in my work was for a “clothing simulation” project.

As a clothing company, producing product photos has always been expensive. In the past, we could only hire models to shoot around 20% of our key styles—time-consuming and costly. In 2023, the general manager of Balabala proposed an idea: skip the models altogether and use flat-lay images, then use AI to simulate how the clothes would look on a model. Today, this approach seems commonplace, but back then, few had tried it.

I still remember the first demo I saw: The flat-lay image was seamlessly “worn” by a virtual model, with results that were both striking and magical. Our business team immediately got excited, imagining that in the future, we could eliminate the need for makeup, styling, and photo shoots.

But the tech team quickly poured cold water on the idea: For standardized items like shoes or cups, the technology worked fine. But clothing presents a much greater challenge—there wasn’t enough training data for the AI models, and the technical team lacked confidence in the feasibility of the approach.

Throughout 2023, we were caught in a tense tug-of-war. Upper management wanted to move forward quickly, but the generated results often looked distorted. We spent a full month testing and refining the solution, working closely with multiple vendors to validate everything from data preparation and model training to A/B testing for visual accuracy. In the end, we concluded that the technology wasn’t mature enough to meet business needs in the short term, and the project had to be shelved.

I wasn’t ready to give up. I turned directly to a business colleague and asked, “Setting aside other factors, if this tool becomes available, would you dare to use it? Would it actually deliver tangible benefits?” After careful consideration, he admitted that, in the short term, he just didn’t feel confident about it.

So we reached a consensus: We couldn’t adopt AI just for the sake of AI. We had to ensure that it would deliver real business value—boosting performance and reducing costs.

Still, this experiment made me realize that AI has the potential to drive exponential productivity gains—and that this is clearly the trend, even if it will take some time to fully materialize. It also laid the groundwork for us to later quickly implement AI-powered solutions like clothing simulation workflows.

In 2024, I first encountered AI Tables—then known as “Multi-dimensional Tables”—at a DingTalk training camp. I was immediately struck by its capabilities: All of a company’s meetings and routine processes could be managed in a single table.

After getting hands-on with the tool, I knew right away how powerful AI Tables were, and I wanted to get my colleagues in other departments using it as soon as possible. One day, I ran into an old coworker in the elevator. During our chat, he mentioned that he was juggling several projects at once. An idea popped into my head: AI Tables could help him. I immediately said, “Let’s set up a time—I’ll walk you through it.”

In the past, we relied heavily on Excel, cramming deadlines and filters into a single spreadsheet and then splitting it among different people. But the data was static, requiring manual updates—and that was not only cumbersome and error-prone but also impossible to keep in sync in real time.

I’ve personally experienced the pitfalls of this approach. One of the most memorable examples involved a product ordering meeting: The clothes were already displayed, price tags were attached—but prices and policies kept changing. If pricing updates weren’t communicated in time, customers’ expectations could diverge from reality, potentially hurting orders.

Back then, we had to fill out countless spreadsheets every day, constantly updating them. We had to create group chats so everyone could update the Excel file in real time. But no matter how we tried, someone always had to manually notify others—and with so many messages flying around, important updates were easily overlooked. One year, a customer complained after placing an order because the policy for a particular style hadn’t been synchronized. A post-mortem revealed that the root cause was “out-of-sync spreadsheet versions.”

Besides product ordering meetings, we also used AI Tables in “time-sensitive” scenarios. The timeline for a clothing product’s lifecycle is tightly controlled: There are strict deadlines for planning, ordering, and launching products, and any delay can disrupt subsequent steps. Previously, we relied entirely on manual monitoring, constantly chasing deadlines. Now, with AI Tables, we can manage everything in one place, automating deadline reminders and notification distribution.

Soon, departments like administration, procurement, the digital center, and the supply chain were all using AI Tables—and the tool was compatible with everyone’s preferred way of filling out forms. Some people love working in spreadsheets; others prefer forms—but in the end, all the data flows into the same table.

A colleague once asked me, “Why aren’t events happening in June grouped together in the calendar?” I explained that the key was shifting our mindset: We couldn’t stick to Excel’s two-dimensional logic anymore; instead, we needed to treat “quarter” as a field. Once he understood this, he instantly saw the light.

Going a step further, we added automated workflows, setting start and end times for tasks along with reminders. In the past, calendars were cluttered with handwritten notes; now, the system automatically sends notifications and closes the loop on task completion. With a single table, we can manage everything from planning to execution.

Looking back, the company’s exploration of AI began with “interest groups.” When new technologies first emerged, there were no standard answers, and our boss encouraged everyone to experiment on their own.

However, he also placed a special expectation on the digital center: As a middle-office department, it needed to take a strategic, resource-coordinating role. He wanted us to bring together scattered efforts and form a company-wide virtual project team.

My role became clear: I needed to build a “bridge” that would help different teams understand AI, adopt it effectively, and make AI tools as universally accessible and easy to use as utilities like water and electricity.

I’ve taken the MBTI test multiple times, and the results have never been consistent. I’m a Gemini—I can be outgoing when the situation calls for it, but I’m equally happy to spend time alone, sipping tea and reading. Some colleagues describe me as a hands-on, hyper-energetic type, while others say I have a very “positive” presence. If my department organizes an event, I’m always eager to participate and help create a lively atmosphere—even if no one else responds for a while, I don’t mind. I’m comfortable with the idea that not everyone is good at everything—and I don’t let that stop me.

This mindset carries over into my work as well.

This March, I traveled to Wenzhou to conduct an AI tools training session. What started as a casual科普 session arranged by the retail training department quickly evolved when I volunteered to include AI Tables in the curriculum. Before long, the word spread: Not only retail managers but also staff from administration, HR, and logistics wanted to attend. The small classroom originally planned for the session couldn’t hold everyone, so we moved to a large conference room that could accommodate over 100 people—and with online participants joining, more than 400 people ended up attending.

A common scenario unfolded afterward: As soon as the training ended, colleagues started reaching out to me. One of the most memorable cases involved Sun Nan—a recent college graduate on a merchandise planning rotation. Her daily work involves collecting size information from “sample officers” and compiling quarterly statistics on new product fittings. After the training, she had a sudden epiphany: This tool could solve her problem! She immediately left a message on DingTalk, asking me to help evaluate its feasibility.

Over the weekend, I created a demo for her, and she was thrilled. “This really works!” she exclaimed. As she dug deeper into the tool, she became increasingly invested, gradually expanding its functionality. We provided her with initial guidance, and she refined the tool based on her specific business needs. As she experimented, the tool grew alongside her.

To help more employees like Sun Nan who want to learn about AI, my team and I created a tool called “Da Sen Tree Hole.” Initially, it was open only to the AIGC group, but as more people expressed interest, we expanded it to the company’s broader internal chat group.

This “Da Sen Tree Hole” is quite interesting. I added an “emotional buffer”: When business colleagues submit requests, if we’re too busy to respond immediately, AI sends an automatic reply first, easing the requester’s anxiety before we take on the task. To date, we’ve received nearly 500 user feedbacks.

After that training session in Wenzhou, some colleagues felt the tools were incredibly useful—but they kept running into problems and frequently asked if I had time to answer their questions over the phone.

With so many inquiries pouring in, Sen Academy reached out to me and asked if I could recommend a “general-purpose” course. I immediately thought of AI Tables and suggested, “Why not schedule another AI Tables training session?”

The resulting three-hour科普session drew an overwhelming response, with more than 300 people signing up.

The impact was immediate: After the training, more departments began adopting AI Tables.

This process has left a deep impression on me: A voluntary training session can bring together those who are eager to learn, and through practical application, AI tools can truly become a frontline productivity driver.

At first, I wasn’t familiar with these tools either. I started by watching DingTalk live streams to learn on my own, and whenever I ran into questions, I turned to the official support team. On weekends, I’d catch up on relevant courses, then connect with business colleagues to address their concerns.

As I’ve conducted more training sessions, I’ve learned that you can’t move too fast—you need to demonstrate even the smallest steps. For example, when creating a new multi-dimensional table, I make sure to explain everything, down to naming conventions, because many people’s habits from Excel don’t apply here at all.

In my self-study, I pay special attention to areas where I tend to get confused, making sure everyone gets extra practice on those tricky parts. Some features seem simple, but many people find themselves “learning it in class and forgetting it as soon as they leave.” If you don’t practice right away, the knowledge fades quickly after the session ends. That’s why I always encourage people to find a partner to practice together before the training begins.

Before each session, I send out a fun questionnaire to get everyone warmed up. Once participants complete it, they can see the changes in real time in the group chat, creating a sense of anticipation. By the time the session starts, everyone already feels engaged and can see firsthand why they need to learn these tools.

After the training, I compile a list of commonly used tools into a dedicated learning hub—a cloud document knowledge base that includes all AI tools relevant to various business processes.

I remember once telling my leader, “To be honest, this work isn’t part of my original job description—but I still feel really happy about it.” My leader paused for a moment, then laughed and said, “You know, when you said that, there was a sparkle in your eyes.”

Today, the company has an AIGC forum group that has grown from a few dozen members to over a thousand. Many colleagues have joined voluntarily, sharing their experiences with AI applications, recommending tools, and learning from each other across different brands. Some outsiders have remarked that the level of engagement in this group rivals that of a professional AI community.

My original motivation for this work has never changed. Back when I was in sales or planning, I was the person responsible for bridging information gaps; now, as I promote AI, I’m still playing the same role. When information flows smoothly, everyone can do the right thing more easily.

My team consists of three colleagues, and I give them a fair amount of autonomy in managing their work. We hold weekly meetings to align on progress, and for the rest of the time, I trust them to move forward independently. Our DingTalk group is called “Boiling Youth,” which sounds full of energy and ambition.

This year, business demands have surged, and we’ve set up multiple virtual teams to handle different projects. Members of these groups come from various departments. I gave one project team the name “Ding San Duo and His Friends”—a subgroup focused on spreading the “AI spark.” I even created a group photo using everyone’s profile pictures as the group’s avatar, to strengthen their sense of belonging.

In the apparel industry, consumer feedback is crucial. Who understands consumers best? Not the designers or the strategy team—but the frontline sales associates. They spend every day in stores, observing how customers shop and listening to their real-time reactions during fittings. Unfortunately, these voices often get distorted as they pass through multiple layers of communication.

In the past, the merchandise department conducted market research only four times a year, traveling across the country. The sample size was limited, and the process was slow. By the time the information reached headquarters, it had already lost much of its value. A customer might say, “The fit is too loose,” but by the time the feedback reaches decision-makers, it may be reduced to a vague statement like “There’s a problem,” leading to unclear decisions.

This is exactly where DingTalk AI tools come into play. Now, when a sales associate casually mentions, “This garment feels a bit tight around the waist,” the comment is instantly uploaded, converted into text by AI, and automatically categorized: Is it a fabric issue? A fit issue? Or just emotional feedback? Different departments can see the information in real time, and the information gap is closed on the spot.

All my convictions stem from a simple principle: Does this initiative create value for the company?

When I first started exploring AI last year, the pace was relatively slow. This year, the situation is completely different. My team and I are being pulled in multiple directions by competing demands, because the business is now truly relying on these tools.

With so many requests coming in, it’s even more important to stay discerning. Sometimes, external sources tout incredible features, while internal colleagues report very different experiences.

Different opinions are inevitable. But when you see positive feedback and enthusiastic support from the business side, you can’t help but feel that this effort is worthwhile—not only because it helps others, but also because it brings a deep sense of personal fulfillment.

Of course, I’m not the most emotionally intelligent person. I often joke with my colleagues that my EQ is “super low.” For example, when my boss assigns me a task, I might bluntly respond, “This is too much pressure.”

Fortunately, the company’s atmosphere is great. Sometimes I tell my boss straight out, “I can’t handle this,” and instead of pressuring me, she sits down with me to sort things out: What are the top priorities right now, and which tasks are truly essential? I often get bogged down in琐碎 details, and she reminds me to prioritize based on value. I’m deeply grateful for that kind of support.

The company’s overall investment in AI is growing. In April this year, Semir’s chairman issued an internal letter, calling on everyone to embrace AI from the top down. Events like AI fairs, AI competitions, and company-wide AI check-ins are being held one after another.

More and more people in the digital center are working on AI-related projects. In the past, our department bore almost the entire burden: We had to secure resources, budgets, and personnel, and we also had to oversee product direction and technological development.

This year, the situation has improved. Someone has taken over the development of the workbench, and the product manager and technical team are each handling their respective responsibilities. I no longer have to shoulder the burden of resources, budgets, and personnel alone. I can focus on科普education and diving deeper into specific use cases. What once felt like scattered efforts is slowly coming together. I can see that I’m getting closer and closer to my goal: truly integrating AI into the way we work.

After three years of working closely with DingTalk AI tools, I’ve come to a deeper realization: “A gentleman is not inherently different—he simply knows how to make good use of the tools at his disposal.” In the end, it’s not AI itself that determines its value—it’s how we use it.

DomTech is DingTalk's officially designated service provider in Macau, specializing in providing DingTalk services to a wide range of customers. If you’d like to learn more about DingTalk platform applications, you can contact our online customer service directly, or reach us by phone at +852 95970612, or by email at cs@dingtalk-macau.com. With a strong development and operations team and extensive market service experience, we can provide you with professional DingTalk solutions and services!