Ten Questions for Zhou Hongyi | After 27 Years as an Entrepreneur, Today Is When I Face the Most "Uncertainty"

AI is moving so fast that your lead might only last a week.

👦🏻 Interview: Koji

🥷 Editing: Yaolan

🧑‍🎨 Layout: NCon

Last night, 360 held a product launch for "Nami Search Agent" at its Beijing headquarters.

After the event, everyone took the elevator up to the building's rooftop for a barbecue gathering.

Koji had a conversation with Hongyi Zhou.

👦🏻 Koji

At today's launch, you mentioned that your Agent chose to run tasks locally in the browser, while Manus went with deploying on cloud virtual machines. Nami chose local deployment for data security and user convenience. How confident are you in this choice now? Will this become the mainstream direction in the future?

👦🏻 Hongyi Zhou

I think every technical approach has its own validity. We have advantages on the client side — after all, we built China's largest browser and know the browser kernel inside out. Local deployment is the most natural fit for us.

Second, many Manus tasks fail in the cloud for two reasons: one, headless browsers lack cookie and user account support, so they keep getting blocked by registration walls; two, using vision capabilities to process tasks in the cloud is prohibitively expensive. Running locally lets us combine with RPA (Robotic Process Automation) and use more "clever" workarounds to improve task success rates.

👦🏻 Koji

A few hours before the launch today, the product UI got a major last-minute overhaul? We were still seeing the old version yesterday, and this new UI feels completely different.

👦🏻 Hongyi Zhou

It is pretty eye-catching. Which UI do you think looks better?

👦🏻 Koji

I think the current interface is very clean. Some Agent products have UIs that feel more like "showing off" — impressive but incomprehensible. But I think Nami's launch UI conveys both power and clarity.

👦🏻 Hongyi Zhou

AI product UIs basically fall into two categories. One is the simple chat interface — weak expressiveness, poor interactivity.

Users don't need flashy. What we pursue is clarity.

The interaction zone and workspace are separated, with cards displaying intermediate results in between, making the process traceable, participatory, and interactive.

Our target users are ordinary people. We want them to clearly see the reasoning process and intermediate results. In the future, designs where you see streaming intermediate results while conversing might become a new trend in interface design. No one can say for sure right now — everyone is still exploring.

👦🏻 Koji

At the launch, you mentioned that even though Agents are powerful now, people often can't think of when to use them. How does Nami plan to solve this? How do you get users to naturally and frequently think of using your agent?

👦🏻 Hongyi Zhou

That's a good question.

Our approach is to stay close to daily life — we don't emphasize professional use cases or deliberately target business applications.

The examples we showed were all about food, entertainment, clothing, housing, transportation — things closer to everyday life with higher usage frequency. But on the other hand, not everyone has deep search needs every day, so we integrate it with traditional search.

Most people still primarily use quick Q&A mode in their daily lives, which is crucial for maintaining user engagement. When users aren't satisfied with results, we can prompt them to try Deep Search or Deep Research — these two modes need to work together. After all, traditional AI Agents often take ten minutes per task, and ordinary users don't have that patience.

👦🏻 Koji

Right, that makes sense.

👦🏻 Hongyi Zhou

Most users don't have that much time.

👦🏻 Koji

So Nami's Agent doesn't ask users to choose "quick search" or "deep search" from the start — instead, the system automatically determines which mode fits better based on the user's query?

👦🏻 Hongyi Zhou

We default to quick search. But after quick search completes, we recommend slow search depending on the situation.

👦🏻 Koji

Currently slow search has ten free uses per day?

👦🏻 Hongyi Zhou

Yes.

👦🏻 Koji

I saw your launch PPT showing eight core features. Honestly, my first thought was — does this mean you haven't found a true "killer feature" yet, which is why you're presenting so many?

👦🏻 Hongyi Zhou

These eight are just technical characteristics, a feature list. It's like building a car — whether it's a sports car, SUV, or pickup truck, that's positioning; but saying the tires are good and the engine is strong, those are features, not positioning. What we need to do now is find a clear, precise product positioning on top of these characteristics.

Because search is fundamentally a very basic, rigid need — users want an accurate answer. Think about it: even with more advanced search forms now, many users still find DeepSeek r1's "thinking process" too verbose.

Some users like multi-turn dialogue and are willing to engage deeply; but more often, users just want to quickly check the weather or look something up without being bothered.

So we believe search should maintain that "search feel" — concise, efficient.

Of course, our first step goes deeper than ordinary search. Because we've layered many vertical search capabilities on top, so while it's still search, it's actually structured intelligent search with more depth.

👦🏻 Koji

Then if you had to capture in one plain sentence what makes Nami Search Agent different from traditional search or competitors, what would that sentence be?

👦🏻 Hongyi Zhou

(Pauses for three seconds, smiles) That's actually pretty hard to come up with. I sometimes struggle with this too.

For example, compared to Lovart and Medeo, our functionality is definitely richer. But if you only do one vertical, like shopping search, that's inherently a long-tail need that can't sustain a complete scenario. So our current thinking is to use "agent search" as a preliminary positioning concept.

Lovart and Medeo both face a problem. The concept of "agent" itself is quite vague. And "vertical agent" — well, it depends how "vertical" you go.

You see, many of our internal agents' capabilities definitely exceed Medeo's, right?

And something like Lovart might break down into sub-domains like graphic design and video design, so there's still a lot of uncertainty in this direction.

👦🏻 Koji

Earlier at the launch, I was flipping through your new book If You Want to Win, You Have to Get in the Game. In the preface, you wrote:

"The real world is like the pony crossing the river. I hope everyone can maintain their youthful spirit and have the courage to test the waters."

But in reality, many people say "I want courage too," yet they just can't find it.

For people in this situation, what would you suggest they do to gradually build up courage?

👦🏻 Hongyi Zhou

I understand what you're saying, but I think there's a bit of a causal mix-up here. Courage doesn't come from nowhere — it doesn't just suddenly appear out of nothing.

There's an old Chinese saying: "A gentleman does not stand under a crumbling wall" — you don't jump into a ditch for no reason. The pony crossed the river because it had a goal. So I believe:

Courage comes from passion, and passion comes from curiosity.

When you truly love something and are genuinely curious about it, you'll naturally want to explore. This exploration itself is purposeful — it's the core driver of your actions. Think about it: if the pony was just running around aimlessly, it wouldn't cross the river at all, and that would miss the whole point of the story.

👦🏻 Koji

So the key is still finding something you truly love and are genuinely curious about.

👦🏻 Hongyi Zhou

Yes, then you'll have very strong internal motivation.

👦🏻 Hongyi Zhou

Back to Nami — we'll figure it out as we go.

The future definitely trends toward personalization and specialization, but the reality is, most people don't have the habit of building knowledge bases. So we're also considering whether to spin out "knowledge base" and hardware products as separate scenarios to explore.

👦🏻 Koji

I actually strongly feel that "figure it out as we go" attitude you described. I think this really is the most appropriate approach right now — because many things don't have clear paths yet, and how technology evolves is uncertain. In this situation, you can only keep moving forward like this.

👦🏻 Hongyi Zhou

But isn't this also the charm of entrepreneurship?

If the answers were already determined and everyone knew what to do, what opportunity would entrepreneurs have?

Because the most certain things are most certain for the giants.

It's precisely because people have different views today that there are opportunities for newcomers.

👦🏻 Koji

You've been an entrepreneur for so many years. Would you say the "uncertainty" you feel right now is the highest, very high, or normal compared to your entire entrepreneurial career?

👦🏻 Hongyi Zhou

The highest.

AI's uncertainty is too strong.

👦🏻 Koji

What does this "highest" mean to you? What does it feel like? Fear? Excitement? Thrill? Or terror?

👦🏻 Hongyi Zhou

Actually, a bit of everything. I'm a pretty complex person myself.

If someone completely doesn't know fear, they actually can't withstand risk. Because they don't sense danger, they won't avoid it, and sooner or later risk will destroy them. But if someone lives in constant fear, they won't dare to innovate either.

So humans are inherently complex creatures. And this uncertainty really is because technology is advancing too fast. Think about it — just keeping up with daily AI progress takes two hours to fully understand. The "lead time" for all products is being compressed extremely short; a lead might only last a week, or even less.

And to put it dramatically, something you just accomplished today might be directly disrupted by new technology tomorrow.

In this situation, you have no choice — you just have to make judgments alone, keep trying, keep trial-and-erroring.

👦🏻 Koji

I remember the day after Manus launched, there was a small communication session. A core concern everyone had was: what's your differentiation? What's your moat? At the time, Hidecloud was very candid and gave a simple but honest answer: "speed of innovation."

👦🏻 Hongyi Zhou

Yes, I've been talking about a concept:

Pessimists are always discussing difficulties, but no matter how much you discuss them, difficulties won't disappear on their own. Optimists look for change in action and build up energy through movement.

Like in judo — have you watched it? When judo athletes are at a strength disadvantage, they rely on agile movement to neutralize their opponent's attacks. Boxing is the same — avoid being in your opponent's center of vision, disrupt their rhythm through constant movement to build your own advantage.

Entrepreneurship is the same — you must keep moving, keep trying. But that doesn't mean moving guarantees success; some people will still fall before the giants.

But you have no choice.

---