Managing $7 billion in annual ad spend, the person who understands global traffic better than anyone just built an AI agent

"What happens when you treat data as a living organism?" By Shi Jiaxiang

Edited by Chen Zhiyan

Li Shuhao may be the person in China who understands global traffic best.

The digital services company Tec-Do (Tec-Do2.0) he founded in 2017 has become one of the most formidable players in China's overseas marketing space, serving nearly 100,000 Chinese companies going global and managing an annual ad budget of $7 billion. Yet as a service provider, Tec-Do has long been accustomed to staying behind the scenes.

Li, now 36, was an early employee at UCWeb. In 2016, cashing out stock from UCWeb's acquisition by Alibaba, he decided to leave and start his own business. After several failed attempts at various tool products, he eventually narrowed his focus to B2B marketing — on the advice of his former boss, Yu Yongfu — which became Tec-Do.

Since its founding in 2017, Tec-Do has completed five funding rounds from prominent investors including IDG Capital, Zhongding Capital, K2VC, and GSR Ventures. He Chuan, partner at Zhongding Capital and lead investor in the Series B, once told us that Tec-Do's core business has "very high barriers." Specifically: first, its scale of ad placement, major media licenses, and ability to serve small clients with AI; second, its coverage of global long-tail media and ability to improve placement efficiency with AI. As the company entered a high-growth flywheel, Tec-Do's revenue grew 5x in four years, and headcount expanded from 300 to 1,500. After 2021, Li stopped opening Tec-Do's doors to capital markets entirely. "The cash on our books exceeds that of many listed companies," he says.

In hindsight, Li and Tec-Do's current position was foreshadowed. After UCWeb merged into Alibaba, Li was put in charge of overseas marketing for Alipay and AliExpress, making him among the earliest Chinese to engage with digital globalization. In the parlance that later became popular, it was undoubtedly a case of "the client-side demoting itself to the vendor side."

Though "globalization" has become the standard answer for Chinese companies seeking growth, market expansion, and incremental opportunity, and though Tec-Do sits squarely in the e-commerce and traffic赛道 that everyone watches, few people know much about the company or Li as a B2B enterprise. Chinese-language coverage of how the company got to where it is remains scant.

Recently, this leading Chinese overseas marketing service provider has made a new move: its first self-developed enterprise-grade, industry-vertical agent focused on overseas marketing, "Navos," is set for release.

Two months ago, in our article "The 'Shadow World' of Globalization", we profiled a series of "shadow companies" occupying pivotal positions in the chain of Chinese enterprises going global — Tec-Do among them.

It was because of that story that, one month before the product launch, An Yong Waves sat down with Li Shuhao for an exclusive, extended conversation. He speaks at roughly double normal speed (apparently at UCWeb, He Xiaopeng constantly told him to slow down). Besides the upcoming Navos, we discussed the full arc of Tec-Do since its founding.

Over nearly three hours, Li spoke about his convictions regarding data and growth, and about how the elitist American traditional marketing world has called Tec-Do "barbarians outside the walls." Unlike most Chinese globalization entrepreneurs, who tend toward restraint, the Shandong native Li repeatedly invoked patriotism, "representing Team China," and his ambition to build a $100 billion company.

This entrepreneur, who hangs the words "data-driven" in large characters on his office wall, follows an intensely pragmatic growth logic in both business development and organizational management. As he himself puts it, he believes in data Darwinism.

The year Li chose to found a B2B company, 2017, was also the inflection point when China's primary-market consumer internet playbook began its decline. Since then, trends have shifted and circumstances changed; B2B companies briefly reached dreamy valuation peaks before falling back, and most investors who once focused on B2B have left the赛道 — even left the investment industry entirely. Li says the reason Tec-Do hasn't died is because it has adapted to environmental changes countless times.

"When the climate turns cold and the forest becomes grassland, those who didn't come down from the trees are still monkeys. Those who came down are humans."

Part 01

A Single Sentence from Yu Yongfu Saved Me

An Yong: You left Alibaba in 2016. Why choose to start a business at that point?

Li Shuhao: When I interviewed at UCWeb, they asked if I wanted to start a company someday. I didn't even think — just said yes, but that I didn't have enough accumulation or credentials yet. Two main reasons for leaving in 2016. First, the stock became sellable. The earnout from UCWeb's sale to Alibaba had matured, so I had some cash and could give it a shot. I actually left a bit earlier than Xiaopeng. He told me not to leave until the stock hit 150, but I sold first anyway.

Second, Alibaba's strategy at the time wasn't focused on going overseas. Cross-border e-commerce was defined as import — helping Chinese people buy foreign products, luxury goods, overseas brands. Back then Alibaba was in a league of its own; internally they said they couldn't find competitors even with a telescope. In 2015, neither Pinduoduo nor ByteDance existed yet, and JD.com hadn't really taken off.

An Yong: At the time, how big did you think you could build the company?

Li Shuhao: It had to be a hundred-billion-scale company, or founding it would be a waste.

I felt internationalization was the ceiling of business. As long as you're not doing business on Mars or the Moon, internationalization is the limit of human commercial cognition.

I also thought I'd discovered a common pattern of large enterprises — grow 3x every year, must triple, absolutely must triple, otherwise I'd be anxious. Only later did I learn that consumer-facing businesses can triple; for B2B services, 50-100% growth is already excellent.

An Yong: Globalization, e-commerce, marketing — did you have this direction clear from the start?

Li Shuhao: When you actually start a business, the early years are extraordinarily anxious.

From 2017 to 2020, society was also quite浮躁. Two types of people dominated: "those casually raising funding in Beijing cafés" — valuations were skyrocketing; and "those getting rich quick at Shenzhen's Huaqiangbei." Friends around me were getting rich, either on paper or in cash.

I'd just left Alibaba with huge ambition, determined to build a major enterprise. But the direction definitely wasn't clear that fast. For the first two years of entrepreneurship, I kept tinkering — site-building tools, tracking tools, probably seven or eight projects at least.

An Yong: How did you eventually narrow down to marketing?

Li Shuhao: K2VC and IDG invested in me early on and were quite tolerant; they didn't interfere much with trial and error. By 2018, Yu Yongfu invested and pushed me to gradually wind down and sell off the other projects through 2019.

Yongfu helped me summarize three forms of B2B services —

The first is intermediary services, making money on information asymmetry.

The second relies on service; you must clarify whether you represent supply or demand. Tec-Do chose to represent the engineer dividend and supply chain dividend — we're on the demand side. Our entire strategy is "serve the client, manage the supplier." This is what Yongfu helped me summarize, and this sentence determined Tec-Do's core positioning. To this day, many B2B companies still haven't figured this out.

The third is building platforms. We don't do traffic monetization; we do data insight and understanding. I don't guarantee any media's sales volume. Because once I do, I'll prioritize selling that thing, and I won't be objective anymore.

I'm not the representative of any overseas traffic source. I only represent the interests of Chinese companies going global.

An Yong: What category contributes the most revenue to Tec-Do now?

Li Shuhao: E-commerce, over 60%. Internally we call ourselves the e-commerce version of AppLovin.

Within e-commerce, apparel is the largest category; consumer electronics is second. But frankly, we don't do consumer electronics well. Consumers have long decision cycles for electronics — hard to drive a purchase from a single piece of marketing content. Same with automobiles; it's not our strength. So there's still plenty of room for competitors.

An Yong: You're not worried?

Li Shuhao: Play to strengths while addressing weaknesses. Prioritize strengths, address weaknesses in parallel. Same logic as ad placement — invest more where ads perform well, keep the underperformers from losing money.

An Yong: Over these ten years, what's the most fatal mistake you've made?

Li Shuhao: From 2019 to 2021, we tried to help clients build the full overseas chain. Going global was too early then; e-commerce wasn't mature. Domestic Taobao sellers didn't even know how to ship. In 2020 we raised two consecutive rounds and invested in site-building, payments, logistics, ERP — even a warehouse. Most of it later failed.

The company expanded from one to two hundred people to four or five hundred. At one point I thought we'd be bankrupt in six months, then cut 200 people.

An Yong: Sounds similar to mistakes from when you first started.

Li Shuhao: Business was growing fast then; it's hard not to get carried away. Felt like what I was doing was right, and it later proved right — it was basically Shopify. But it wasn't what our company should have been doing at the time. Think about it: what gave me the right to handle "ERP + site-building + payments + logistics + marketing" all at once? I didn't have that capability. But when we reach sufficient scale in the future, I still want to do this.

Part 02

"Representing Team China"

An Yong: This is your ninth year of entrepreneurship. What's the team's current scale?

Li Shuhao: We now have about 1,500 people, serving nearly 100,000 clients, with coverage of over 80% of top-tier overseas clients.

An Yong: Compared to business scale, your domestic visibility and image seem somewhat excessively low-key.

Li Shuhao: Actually, our main changes happened in the past three years. The data system was built out by 2020, but on the content side — creative assets, for instance — I'm completely unskilled, and couldn't take clients to find local celebrities in the US like a 4A agency. But because of AIGC, I've used AI to supplement content: secondary creation editing, digital humans.

This is the biggest gap we've closed in the past three years, which is why revenue scale has risen sharply over four years.

An Yong: What magnitude of increase?

Li Shuhao: 5x in 4 years.

An Yong: That much?

Li Shuhao: Not that much for our industry — Pinduoduo did 10x in two years.

Two critical junctures brought us to where we are:

First, breaking through the limitations of ad placement and user reach through technology.

Second, using large language models, Stable Diffusion-powered text-to-image, and text-to-video AI content creation to flatten cultural differences.

The critical juncture we're now passing is AI's understanding and insight into the world. Each breakthrough at these key nodes brings two to three times efficiency gains.

An Yong: What do you most want to do now?

Li Shuhao: Last year I shouted a slogan internally: "Represent Team China." Why? So that overseas companies don't look down on Chinese enterprises. We understand e-commerce, we understand digital, we understand marketing — we're no worse than them. In the entire traffic industry, are Chinese really worse players than Americans?

I deeply admire ByteDance because it's a company with very successful internationalization. Why doesn't Yiming Zhang take it public? When it goes public, it has to live up to his name — make a stunning impact.

An Yong: Among overseas founders today, we rarely hear such views. Most prefer to blur their positioning.

Li Shuhao: A very important reason I left to start my own business in 2016 was disappointment with Alibaba's internationalization strategy at the time. Because I didn't want to help foreigners earn Chinese people's money — I wanted to help Chinese people earn foreigners' money. This might be related to me being Shandongese; I have some patriotism. Earning foreigners' money — I feel quite proud of that.

Once, a boss told me: when you start a business, think about whether you can tell your son about your life experience? Can you make your mother proud? I said: that makes sense. I translated this as — can you become a company people respect? That really moved me.

An Yong: You've mentioned several times that you're Shandongese, but going to start a business doesn't seem like a typical Shandong person's choice.

Li Shuhao: Shandong people have many characteristics; I think an important one is this sense of patriotism. If foreigners can do something, we can do it too, even better — this is something that boosts Chinese morale.

An Yong: For building a global enterprise from China today, is this patriotism you speak of a necessity?

Li Shuhao: Not a necessity, but it makes it easier for a company to survive cycles.

Part 03

Using an AI Agent to "Decentralize" Marketing

An Yong: What's the primary purpose of the AI agent product Navos you're launching?

Li Shuhao: It's an autonomous driving system for overseas marketing. It understands all the steps and problems in the overseas marketing process. Strategically, it can analyze each country's characteristics, whether there's demand for your product, what its selling points are — this is the brain. At the same time, overseas marketing is also a very fragmented engineering task: you need to create massive amounts of ad assets, run hundreds of accounts. Humans doing this would be overwhelmed, but Navos handles it easily.

An Yong: Compared to numerous AI marketing companies, what's Tec-Do's core competitive advantage in building Navos?

Li Shuhao: First, benchmark-forming data — something I've been accumulating since founding the company. For very general marketing questions: what sells in women's clothing? What assets perform best? Or for halal beauty in Indonesia, what's the average conversion cycle and customer price? Other companies don't have the data, so they don't know. But I've accumulated hundreds of millions of assets and billions of strategies annually over the past years; without this volume of data, you can't train it.

The second barrier is feedback. How should creative ads be optimized? Where's the direction for secondary training? Most content needs scoring. At Tec-Do, the people scoring manage nearly 10 billion in ads annually — completely different from other agent tools having consumers score.

I believe there will be a year of chaos for agents. Right now everyone's agents look very similar; then it becomes whose data has production-guiding significance. Historically, the best technologies have always been adopted by marketing first — recommendation algorithms, search engines. Marketing is the place closest to commercialization; without accumulation, it's hard to do.

An Yong: What kind of customers do you expect to use it?

Li Shuhao: First, beginners who want to make TikTok work — through Navos they can learn how to sell, how to select products, who sells best, how to sell, to whom. But what to do next step, I won't manage.

Second, existing or target clients with clear demands. For example, a client wants to reach top 5 in cosmetics. I have a benchmark — telling them what the top 10's sales volumes are, what their ROI is. If they have existing raw assets, just tell me and I'll help with secondary creation, content understanding. For distribution, Facebook or Google?

An Yong: That's now; who will this agent product be sold to in the future?

Li Shuhao: To all creators who want to do business. Every creator should be a company. We want to decentralize marketing; content must belong to creators.

An Yong: What changes do you expect Navos to bring to Tec-Do in the future?

Li Shuhao: We hope this marks the beginning of Tec-Do 3.0. Many people don't know why our company is named Tec-Do2.0 — actually we want to build a company like building a product, constantly iterating and upgrading.

Let professionals do professional things; leave marketing to Navos.

An Yong: Tec-Do is also making investments in the AI space. What projects have you invested in this year?

Li Shuhao: 2025 is the first year of AI applications, just as we considered 2020 the first year of going global. But I haven't invested in a single agent project, because many agent products today don't even have a use case — it's crazy. As a corporate VC inside the industry, we shouldn't play along with VCs. We're not envious of these people at all.

But we've invested in seven or eight AIGC companies, because content creativity is never finished. I believe this wave makes grassroots entrepreneurship very difficult; content creativity is still a bottom-up thing, but agents are top-down — without data accumulation, it's very hard.

Part 04

A Believer in Data Darwinism

An Yong: Before AIGC arrived, what do you think was Tec-Do's core differentiating competitive advantage that enabled it to reach its current scale?

Li Shuhao: BI (business intelligence — helping enterprises transform massive data into actionable commercial insight). We're always capturing technology dividends. At the time, Chinese companies overseas had no data; I just needed to organize the data to do this well.

I rarely call myself a marketing company. We're not the traditional 4A advertising agency model. I'm just helping you optimize data, helping you understand your users.

An Yong: What's the essential difference from traditional marketing?

Li Shuhao: It's like the difference between "Shennong tasting a hundred herbs" and "God creating humans."

Those international 4A companies are "God creating humans" — wildly creating a big creative concept that enters the client's flow. They don't even know that ordinary people living in Los Angeles might spend five dollars on a piece of clothing, two dollars on a T-shirt. The 4A elites have never seen this kind of life.

We are Shennong tasting a hundred herbs. Our testing style is "everything not forbidden is permitted." You set the bottom line; leave everything else to data. Many assets that look terrible at first glance are worth trying. 90% of assets perform mediocrely, but very likely the remaining 10% that look ordinary suddenly explode. They spend $1 million on one creative; I spend $100 on one creative and test 10,000 — how could I not find one better than yours?

Don't start with 5 experts, 10 experts deliberating on what tone our game should have. There's no tone. Consumers clicking in is the tone. They think we have no class; we think they have bad data. So we're fundamentally not the same kind of people.

The Americans say we're just barbarians outside the walls. I say, "We're going to use our strength to rewrite the old order and logic."

An Yong: Why do you believe so deeply in the power of data?

Li Shuhao: This is the "data Darwinism" I learned at UCWeb and Alibaba: what survives is what's good. I call it "data Darwinism."

UC Browser grew very fast in India and Indonesia. Many analyzed that it was because network infrastructure was poor in those countries, with many users, and UC had deployed localized servers so Indian users didn't experience lag. In reality, we just pushed through data. We promoted globally, and the data showed India and Indonesia performed best.

2 billion consumers see our ads daily — this is the collective intelligence of 100,000 clients. What gives you the right to say you understand better than everyone else? We firmly oppose all preemptive behavior, all self-styled prophecy.

We make so many strategies, yet we can't tell you whether your product suits 40-year-olds or 50-year-olds. All our predictions draw conclusions from massive data and reduce your trial-and-error cost — not eliminate trial-and-error.

A product's features will never find the correct answer before testing; only testing reveals it. Any product, given sufficient promotion, will inevitably be selected by consumers.

An Yong: Do you also rely heavily on data for company management?

Li Shuhao: Exactly. I originally had 10 BI products internally; internal evaluations all used BI. Only data speaks. Even the brand department had to work with data, so brand people have a hard time here.

Two years ago I had some spare money and asked a master calligrapher to write something. Others get "Grand Ambitions" or "Great Virtue Carries All." I asked the master: could you write "Data-Driven" for me? The master didn't understand — "What is data-driven?" — refused. I eventually convinced him with effort.

Now this "Data-Driven" hangs in my office.

An Yong: After Navos launches, will you want to be defined as an AI company?

Li Shuhao: Fundamentally still a BI company. Though capital markets lost their fascination with BI long ago. I don't care; BI must be emphasized. Data is always the most stable.

Several friends around me who work with data share this near-mad obsession. We treat every piece of information in the data as a living organism: it wants to survive on its own, it's reproducing and evolving.

Mendel observed fruit flies over days; I can see changes every day. This is the lowest cost with the fastest feedback.

An Yong: If feedback comes faster and faster, won't your threshold keep rising?

Li Shuhao: So I can't accept people who don't progress, but I can accept people who aren't great yet. Competition throughout internet history has always been "survival of the remaining" — the remaining剩, not the victorious 胜.

Every year or two I post new reflections on entrepreneurship in my Moments. Later I realized these reflections all mean the same thing: the reason we are where we are is because we've adapted to environmental changes again. When the climate turns cold and the forest becomes grassland, those who didn't come down from the trees are still monkeys; those who came down are humans.

This process brings me joy. Site clusters gone, we're still here — I'm happy. APPs gone, we're still here — I'm happy. Marketing information-asymmetry players gone, we're still here — I'm happy. Influencers and top KOLs gone, we're still here — I'm happy.

An Yong: What makes you most anxious today?

Li Shuhao: What I agonize over daily is: look at the best American AI company product launches, the Chinese scientists in them aren't very old. The AI era changes so fast — whether machine learning or recommendation algorithms, the people achieving the most now aren't necessarily those who'll achieve the most in the future. I'm uncertain how to choose talent.

This is also caused by AI's underlying technology innovation changing too fast — every six months, even every three months, large language models, pre-training, agents update. I'm also anxious about the speed of embracing change. You call it anxiety, but it's also excitement — it's because of these things that there's opportunity to build an even more awesome company.

An Yong: Does Tec-Do's current high growth have side effects?

Li Shuhao: I could say many correct things, but honestly — I've studied Huawei, studied ByteDance. Growth can cover all problems, though it does have certain side effects.

Internally we also have a term: "钝感力" (thick-skinned resilience /钝感力). We advocate钝感力. We know what problems exist, but we must keep growing.

I believe everything unrelated to growth can be钝感.

I know high growth will inevitably bring difficulties, but I persist in this. Continuous growth with continuously satisfied clients — let "钝感力" solve all other problems.

An Yong: Growth curves can't stay steep forever; growth itself can't be eternal.

Li Shuhao: In 2020 I set my entrepreneurial goal: survive 20 years first, ensuring my energy and health hold up, working at least until 50. Chinese civilization is 5,000 years old; wheat has ripened 5,000 times. Our iteration cycles far exceed 5,000. That's already very happy, very fulfilling.

An Yong: You seem to like using biological laws as analogies.

Li Shuhao: Because several friends and I enjoy chatting this way, and it fits how I understand the world. It's the most visceral way, letting me understand the world at the lowest cost.

Image source | Unsplash

Recommended Reading

Where Money Flows, People Rise and Fall