From Selling Robot Vacuums to Billions, a Post-'95 Entrepreneur Enters the AI Hardware Race | Z Talk

Z Talk is ZhenFund's column for sharing insights.

Recently, Lexiang Technology, founded by former Dreame China executive president Guo Renjie, announced the completion of a nearly RMB 200 million angel round at a valuation of approximately RMB 600 million. ZhenFund participated in the investment.

Born in 1997, Guo entered Xi'an Jiaotong University's youth class at age 15. In early 2021, he joined Dreame to manage a team of over 1,500 people, leading Dreame's China region to grow from zero to an annual scale of RMB 6 billion.

Below is the full interview between Guo Renjie and AIEmergence.

By Deng Yongyi | Source: AIEmergence (ID: AIEmergence)

How long does it take to build a company's GMV to over RMB 10 billion?

Guo Renjie, now 27, did it in just three years. In 2021, after 11 months as a management trainee at Procter & Gamble, he joined Dreame, a startup that had only been founded four years prior.

Guo was lucky, but he also had the ability to seize that luck. At the time, Dreame had just begun its transformation from OEM manufacturing to focusing on its own brands. Three years after joining, Guo helped Dreame achieve annual domestic sales exceeding RMB 10 billion, making him a well-known operator in the consumer electronics circle.

To outsiders, this looks like a smooth path of a "king of grinding": entering university at 15, completing graduate studies at 21, earning a million annual salary at 23, and by 26 becoming the China executive president of a prominent new consumer brand.

But in December 2024, Guo made a decision: to end this story of early fame, tear everything down, and start over.

"I'm 27 now, a so-called China executive president, but what comes after? Who do I become at 29, at 31? But if I go out and start a company, today's story ends. I get to define a fresh start."

AIEmergence has learned exclusively that "Lexiang Technology" recently completed its angel round, led by IDG Capital with ZhenFund's participation, raising nearly RMB 200 million at a post-money valuation of approximately RMB 600 million. Following this round, Lexiang Technology will continue investing in R&D for its first product, as well as team building and technology reserves.

The choice of AI hardware stems from Guo's assessment of his own strengths: finding AI hardware consumer products with sufficient certainty and proven product-market fit — suitable for rapid zero-to-one scaling.

Over the past year, market disappointment with large language models may only be one slice of the picture. If you look toward manufacturing hubs like Shenzhen and Hangzhou, or the recently concluded CES across the Pacific, you might see a rather different scene: robots, AI glasses, and other categories have entered a "hundred schools contending" phase, with some already generating sales exceeding RMB 100 million. Meanwhile, novel products like AI toys, AI bird-watching devices, and AI walking sticks have become dizzyingly abundant.

Compared to large models and AI applications still shrouded in fog, hardware is a domain and battlefield more familiar to Chinese players. AI hardware entrepreneurs also generally harbor ambitions to "surpass Silicon Valley and lead the world."

This confidence also stems from the hardware export wave of recent years, in which Chinese manufacturing has completed a "refresh" of its image.

As Guo began his new project, we spoke with him. This story is about how a young man seized an era's dividends, then put everything aside to plunge into an uncertain new wave with a new posture.

Guo Renjie's background is exactly the kind investors in today's venture capital circle favor — young, having spent big money and won big battles, wild enough, and restless enough. He also demonstrates a more aggressive entrepreneurial style: whoever does well, go surpass them.

For his first startup, his goal is simple: choose a thick enough ski slope, so that "even if I deviate a bit, I won't slide off" — survive first.

He also makes no secret of his ambition: "My core competency is grinding."

---

Children Are the True AI Native Users

Q: In 2024, from when people first rumored you might leave to when you officially announced, it was only about two months. When did the idea of starting a company take shape?

Guo Renjie: Around August 2024, I think. Once I decided, I went to talk to CEO Yu (Dreame's CEO).

Actually, before joining Dreame, I had expressed to CEO Yu that I wanted to experience the process of entrepreneurship, which is why I wanted to join Dreame and build together. In 2024, I reached a milestone — Dreame became number one in the industry. I had achieved what I wanted to achieve, and felt ready for the next step.

Q: Did you hesitate? Leaving Dreame at this point — they're almost going public.

Guo Renjie: Not for a moment.

I had two reasons in my mind. First, I wanted to experience the process of entrepreneurship. I had done reasonably well at Dreame, so it was time to go.

Second, I wanted to break the predetermined path of "15 years old entering university, 21 graduating with a master's, 23 with a million annual salary." I needed the courage to break and rebuild, otherwise I would keep being carried forward on this track.

I'm 27 and China executive president — what next? Who do I become at 29, at 31? If I go out and start a company, today's story ends, and I get to define a fresh start.

Now I'm just a small boss of a company with nine employees. Nothing particularly worth showing off. Just focus on doing what I need to do well.

Q: What have you been doing these past few months since leaving?

Guo Renjie: After leaving, it took a week or two to clarify direction, then I started working on this funding round.

Q: Lexiang chose the AI hardware direction. How did you make this judgment?

Guo Renjie: I have several criteria for choosing a track: First, no absolute innovation. I only want to do product upgrades within basically validated categories. So AI software — large models, cutting-edge robots — those definitely have nothing to do with me, because those tracks are still in technology investment stages.

Second, must do global markets — domestic R&D, overseas sales.

Third, hopefully something with AI and robotics integration. This is partly because I'm relatively young, so my understanding of new technologies is likely faster, and it better matches consumer product cognition.

It's simple: if you're purely technologically ahead, you might get copied in three months. But if integrated with new technologies, maybe you stay ahead for five months. Those extra two months are crucial to me. Because I rely on a grinding business model — with those two months, I can amplify my moat.

Based on these criteria, I looked for tracks in AI hardware that have achieved PMF (Product Market Fit) and can address global markets.

Q: I heard you're making products for the children's market?

Guo Renjie: I want to do everything, but I always feel children are the true AI native users. If doing AI hardware, there's a logic to start thinking from children's perspective for inspiration, though the final product may not only be for children.

If I made hardware for you, choosing between your phone and this hardware, you'd probably choose your phone. What hardware could be more fun than a phone? When we chat with Doubao now, it's mostly with a teasing mindset, because we think of it as just a model.

So this is a great AI product user profile: children are curious about AI products, don't have phones yet, and every mother's desire is to pull their kids away from phones, replacing them with other physical devices.

Q: Children's products also have slightly higher tolerance for error.

Guo Renjie: You're absolutely right. If I made hardware for you, you'd play with it twice and throw it away, then post on Xiaohongshu raging at me. But if a child plays with something twice and stops, parents can still accept that.

If facing such users, you still can't make an AI robot hardware, and you say you can do the adult market — I don't believe it.

Q: But hardware also has different types — emotional companionship, or more educational. What do you envision your product to be like?

Guo Renjie: Playability is the first aspect I focus on.

One of my thoughts is that at this stage, the core moat of AI hardware products still lies in hardware interaction, not underlying software. I think software will eventually merge with hardware.

There are several clear cases. Take Tesla's robot — everyone thinks it changed the world. But I think at the software foundation level, the difference between it and domestic players like Galaxy General Intelligence, Zhiyuan, and Unitree may not actually be that large.

But Tesla did one thing particularly well: interaction. It's more human-like — expressions, micro-movements — so people feel it's more advanced, naturally more intimate.

Q: If you want AI hardware to behave more human-like, doesn't that also depend on software?

Guo Renjie: I think the first step should still be hardware. For example, if a robot has no mouth and relies on a rear speaker for sound, versus a robot that opens its mouth when speaking — for a child, the experience is definitely different.

Hardware interaction is still the first step. To do interaction well, the hardware barrier is also relatively higher.

Q: So this industry is still in a hardware-driven stage?

Guo Renjie: Currently, most AI hardware software is externally connected. I think the most important thing from 0 to 1 is first operations — survive first, get scale and profit. After that, I'll definitely lay out some moats, like self-developed motors, LiDAR, and eventually model software.

Q: How do you understand moats?

Guo Renjie: Moats only depend on time difference. Today you can do it, others can only do it in three months — you earn the premium of those three months. Once they catch up, you find the next thing.

I don't really judge what's a "dividend." I only judge whether, through efficient iteration in an industry, I can open a gap. If I can open it, I can do it.

Dividends are a subset of opportunities. After judging that something has opportunity, then you look for dividends — that's the most efficient approach.

Q: What's your assessment of the current AI hardware market?

Guo Renjie: Very early, very primitive. I recently saw Anker talk about its "shallow sea strategy" (attempting to expand through more细分 categories to become a more heavyweight company), which really resonated with me.

I think I'm now looking for "beach strategy" products. What's a beach? It's where everyone comes to take a look. Big companies are like waves — they come up, take a look, and leave.

Because they'll feel this thing is too new, at most assigning a five-person small team to study it. That leaves space for small players — I'm one of those small players.

Q: Do investors have doubts about this track's ceiling?

Guo Renjie: I believe this is a long-term correct, big ski slope. A slope big enough means even if I have setbacks, I won't be eliminated.

Does this track have value, can it birth strong companies? Definitely has potential. If everyone can find PMF in this track, it's like picking shells on the beach — stringing them together is also valuable.

---

My Core Competency Is Grinding

Q: Many people are curious about your career path: studied energy and finance in university, then dove headfirst into consumer goods. 11 months at P&G, immediately went to Dreame, became China president in three years — a very "winner in life" path.

Guo Renjie: But for my background, this wasn't a typical choice.

I originally studied engineering in university, liked tinkering with practical things. Second, when doing my finance graduate studies, I realized I wasn't interested in these virtual numbers. I care more about things that can generate actual impact, that I can control myself. Putting it together, the goal was clear: "I want to do industry, I want to start a company."

I asked around, and everyone said the simplest way to understand industry is to go to P&G.

Q: But you were at P&G less than a year before going to Dreame. What did you do there?

Guo Renjie: I was at P&G for only a year, mainly did two things. First was rotation — from Marketing to Sales, then did something like product manager work for a while.

I tried to stay in as many different departments as possible, like marketing, sales. For other departments, I used the method of treating people to meals to understand — found seven or eight people each in R&D, procurement, IT, and other departments to eat with, asking what they were doing. After several months, I figured out how an industrial business operates.

Q: You were still very young when you went to Dreame, and Yu Hao (Dreame CEO) let you lead the marketing team from the start. What convinced him?

Guo Renjie: Everyone was young then — I was 23, CEO Yu was 10 years older. At the end of the interview, I asked him two questions. The first was: "I'm so young, why do you dare to use me?"

He said something like "the world is a草台班子" — basically: the world isn't as complex as it looks, and no one is really that amazing. If a young person is twice as smart and iterates five times as fast, that's ten times the efficiency.

He also said, never having worked might not be good, but might not be bad either. Because without experience, we dare to try, constantly trial-and-error to find the right path, rather than judging a right path through experience and logic.

Q: But resources are limited at any given time.

Guo Renjie: Resources are definitely insufficient, but our core is making everyone dare to admit mistakes — the cost of admitting mistakes is very small.

Others iterate once, I iterate ten times. We ensure that the cost of us admitting ten mistakes is the same as others making one mistake, so we're definitely more accurate. Then grab all the dividends, and you'll definitely do better.

Q: How much patience do you generally give new projects?

Guo Renjie: Our minimum trial-and-error unit is really small — we might give people RMB 10,000 or 20,000 to test new products. On TikTok, you can test whether a product is good in a week.

Most of the time, you know in the first second whether something is right or wrong. But many corporate managers in big companies can't admit they're wrong — they might end up spending RMB 5 million before daring to say this thing was wrong.

Q: This is in the hardware industry — hardware's design-to-manufacturing cycle is much longer than software's.

Guo Renjie: This is also an inertia mindset — people think the hardware industry can't be tested. So we push testing earlier, like starting pre-sales at the prototype stage. Once data looks bad, quickly switch products.

Many outsiders see us expanding into many categories then quickly abandoning them, thinking we're burning money randomly, ignoring our精细化 work. But this is precisely the key to our success.

Q: You became Dreame China president in three years. What did you do right?

Guo Renjie: From day one, I agreed with CEO Yu's logic — grind and iterate like crazy, and trial-and-error like crazy.

Q: Would anyone disagree?

Guo Renjie: For many people who've worked for years, it's not that they disagree — they can't accept it.

Many startups think their core is technological innovation, business model, or capturing dividends.

I think many startups grind without realizing why they're grinding, just under pressure. But we never have. I think grinding is my core. But we treat grinding as a method, as the only weapon we have.

Q: How much grinding, give an example?

Guo Renjie: For example, for the same piece of content, we might iterate at least 3 to 5 times a day, while competitors might do it once.

We also build high-elimination, high-incentive organizations — higher incentives for hosts, operations, field control. Competitors end broadcast at 10 PM, our people spontaneously broadcast later.

Q: What important battles did you experience at Dreame?

Guo Renjie: TikTok startup counts as one — we went from 0 to nearly RMB 30 million in three months. Actually, I shouldn't have been doing TikTok then, since I was only responsible for marketing. I purely felt TikTok was too important not to do.

Q: What was TikTok's state then, why so confident?

Guo Renjie: Starting up on traditional channels like Tmall and JD.com might take three to five years, but TikTok was a new market. So this had to be done.

At the time, TikTok was transitioning from white-label to brands — this was the same stage Dreame was at. If we could ride this momentum, we'd definitely rise.

But fundamentally, it was believing TikTok would rise. I think this relates to us being young. Because by then, I rarely actively thought about buying something — I'd see something interesting and buy it. We felt "product finding people" was the future shopping scenario.

A comrade and I made a plan, roughly calculated what the current industry leader's level was, asked CEO Yu for roughly RMB 8 million budget, promising to become industry leader in three months.

Q: Did you feel confident then?

Guo Renjie: No confidence, but just did it.

Q: What if you failed?

Guo Renjie: Anything was fine, even getting fired I could accept. Because I was young, no burden — worst case, I'd leave.

I came to learn entrepreneurship. I wouldn't think about whether I should do this thing — I just had to do it well. I wanted crazy growth and experience in this company. Good or bad, I'd accept it, be happy with it.

Q: How did you grind to success in the end?

Guo Renjie: Initially there really were no people, I didn't understand anything. Later we organized through high incentives, high elimination, internal grinding — found many young people with learning ability, learned everything from 0 to 1 ourselves, and grew that way.

Later it divided into roughly three steps. First, learning from competitors to invest in livestream information flow, but found new brand livestreams had no viewers.

Then started doing content jumping to livestreams. We were the first in this industry forced to do content strategy: first invest in short videos, users get interested, enter livestream through clicking avatar.

At first it was imitating爆款 content, because those were scripts others had validated. Second,强制 requiring every TikTok tag to have our related videos,蹭 all hot spots, so natural conversion probability would be high. Doing these two points completed the startup from 0 to RMB 100,000 GMV.

Q: So overall it was still about doing your best at every node.

Guo Renjie: The real explosive node was still through testing all product selling points, seeing which selling point or product could explode. Like permutations and combinations — using products to absorb water, milk, cat litter, dog litter, cat food, dog food — finally discovering wet-dry vacuum + absorbing instant noodles could explode. One video achieved RMB 1 million GMV in a day, and TikTok truly started allocating traffic to us exponentially.

Many online say I invented "absorbing instant noodles" — I can't understand how it's blown so mysteriously. This was also tested out by the team together, not some new invention.

Q: From your perspective, how do you evaluate whether business is doing well or poorly?

Guo Renjie: Only do first place. Every position's team member needs to explain to their leader how to measure whether they've achieved industry first in this position. If you're responsible for KOLs, then look at whether your product's share with each KOL is industry first; then we look at specific business data.

Q: "Be first" is something you often mention. Why?

Guo Renjie: "Be first" is the most efficient management tool. If the industry leader sold RMB 1 billion, and you decide to be industry second, what's the target? RMB 900 million? RMB 800 million? By the time you discuss it, a quarter might be gone. But being first is simple — I just need to hit RMB 1.01 billion.

Second, we believe if every business single point achieves first, the overall is first. We never discuss how to better combine Department A with Department B — just give me single-point first, even if there's redundancy and internal friction in between, we don't care.

---

The Future of AI Hardware Is Still in China

Q: The years Dreame grew were also years of domestic brand innovation. Where did your obsession with becoming "world number one" come from?

Guo Renjie: I strongly agree with a point CEO Yu made: the previous generation of Fortune 500 manufacturing companies was divided by R&D, supply chain, and brand platform. R&D in America, supply chain in China, brand design in Europe. This was the traditional manufacturing world order — Apple, Samsung are all like this.

But new-generation manufacturing has changed — R&D has returned to China. Because we have the highest quality and highest quantity of engineers. Electric vehicles, drones, phones — the R&D and manufacturing of these emerging products have come back. Traditional fields like chips and pharmaceuticals coming back is only a matter of time.

Q: I feel things have changed these two years — people are much more confident about domestic brands.

Guo Renjie: This also needs to seize dividends early.

For example, when we first entered Bilibili to sell robot vacuums, we suffered for a while. TikTok was hard for robot vacuums because navigation demonstrations take a minute, and people basically couldn't sit through such long videos. But all companies felt Bilibili users had no money, wouldn't buy RMB several-thousand robot vacuums.

But I was young! I was a Bilibili Level 5 annual member. Hearing this view, I went all-in. Precisely because everyone thought this way, we felt Bilibili definitely had dividends. Can otaku people buy RMB 3,000 figurines but not robot vacuums? Now Bilibili has many young people who deeply love Chinese tech companies and support domestic products.

The information gap was this large. We later quickly cooperated on one long video on Bilibili — it exploded, one video sold RMB 30 million GMV. ROI was constantly 1:8, 1:9 — completely unimaginable.

Q: Now that you're in AI hardware, if it's such an early industry, is this grinding approach still applicable?

Guo Renjie: Applicable. I think the current market is in the 0 to 1, 1 to 10 stage — I have to rely on this approach. In stages with PMF, with target categories and anchor companies, I can quickly grind and achieve surpassing.

Q: What's this year's goal?

Guo Renjie: Recently, probably first spend a month or two converging on product direction, doing market validation, hoping to produce products with scale and monthly profit in the first year.

Q: What do you think your strengths are?

Guo Renjie: I think companies divide into two types. First is product manager-driven — their core is polishing products well, they love this thing, making the product the best.

Second is determined by business model — stronger in operations, sales capability, product planning ability, able to find some opportunities and amplify them.

I definitely belong to the latter. I can grind — when I compete with you, I'll definitely out-grind you.

Second, based on existing products, I can find some upgrade and innovation points.

When we did wet-dry vacuums at Dreame, the category already had Tineco. We were actually the second wave of brands, but we made it the biggest. We have relatively strong user insight capability, so we can find some market gaps and build them up.

Third, because I'm young, my understanding of new things, new technologies, new channels is relatively faster. This is also why we could capture some dividends before.

Q: What's the question you're most often challenged on?

Guo Renjie: People feel my life has been too smooth, that a young person this smooth will definitely hit pitfalls. But actually, I also experienced many career ups and downs at Dreame — people just don't talk about them. But my resilience and mentality are both very good.

Q: Where does your underlying motivation for decision-making come from? You keep saying you want to start a company, want to do things with greater impact. How do you understand the word "impact"?

Guo Renjie: I want to create value, and second, help others.

I went to university at Xi'an Jiaotong University. Many people probably can't imagine how severe the information gap was between us and classmates from developed coastal cities. Everyone studied very solidly, but nearing graduation still didn't know the social operating rules of big cities, didn't know that to go into finance, consulting, and other industries you needed internships.

So I founded a career development club at that time, organized outstanding seniors from various industries across years to come back and give lectures, telling everyone how investment banking, consulting, insurance, and other industries operate, and how to optimize resumes to get these jobs.

I did this completely free at the time, completely couldn't think about commercialization or making money — just liked leading everyone together, breaking information gaps, creating value for everyone.

Since childhood, I've experienced some major family changes, switched in very extreme environments, had plenty of reasons to go bad. But many people helped me then, letting me grow into who I am today, so I want to help more people — that's life's true value.

I hope to leave some of my views and ideas in this world. Only with sufficient success can I have the resources — money or influence — to help more people.

This article was written by Deng Yongyi, first published on the WeChat public account "AIEmergence" (ID: AIEmergence), covering the emerging industrial revolution in the new era, an account under 36Kr.

---

Recommended Reading