Conviction, Patience, and Confidence: Lessons from a Decade of Smart Glasses Ups and Downs | Linear Voice

2025 is the breakout year for AI glasses. With major tech companies rushing to announce their entry and startups spinning dazzling new narratives, more people are shifting from merely "knowing about" smart glasses to actually "wanting to play with them" and "wanting to buy them."

Amid this wave, Rokid stands out as a pioneering startup that has survived multiple cycles and remains at center stage. Regardless of how market winds shift, its team has spent ten years obsessing over one thing: how to turn a pair of glasses into truly good smart glasses. Linear Capital, as Rokid's earliest-round investor, has witnessed every up and down along the way.

Today's content comes from a talk by Zhao Weiqi, Rokid's Global Open Ecosystem Lead, at a Linear Capital "Tech Pi" event. He broke down Rokid's underlying thinking for surviving a decade-long cycle through three dimensions: original intention, patience, and conviction. In his view, original intention is the compass that points the way; patience adds depth to dreams and products; and conviction comes from the experience of crossing cycles. No industry ever advances through one company alone — it requires many participants moving forward together.

In the wave of AI hardware entrepreneurship, people have witnessed countless products rise and fall. Standing at the inflection point where AI and hardware are accelerating their fusion, how do founders survive technological cycles and market volatility? When major players enter and ecosystems get reshuffled, how can startups find their survival gaps and long-term value?

Rokid's decade of practice in crossing cycles seems to offer some answers that return to fundamentals: anchor direction with original intention, dig deep into the industry with patience, and cross the entrepreneurial cycle with conviction.

Hardware competition ultimately comes down to deep understanding of human nature and persistent commitment to engineering long-termism. The real moat lies not in chasing trends, but in whether you can soberly answer amid the noise: Who are we building for? What problem are we actually solving?

From PC to mobile internet, to spatial computing, and now to the AI era, you can see that human-computer interaction has been moving in one direction: the interconnection of everything.

I was fortunate to experience this evolution firsthand. I studied mathematics, started in software, and was among the earliest developers of Alexa, participating in smart speaker development. Later I expanded into spatial computing, founding an AR extreme sports company in Silicon Valley. As a hardcore geek for multimodal AI software and hardware, I joined Rokid in its early days, responsible for the system and full software stack of the first-generation AR glasses. I then went on to co-build and scale core business units, and began leading the construction and expansion of our global open ecosystem.

Looking back at the direction of technological migration, it's a process of computing power moving closer to people, of machines understanding humans better.

From studying mathematics, I've always wanted computers to help people unlock more of their capabilities, freeing up more time and energy to enjoy life. For many entrepreneurs, when you first assemble a team or conceive an idea, you always start with an original intention. Perhaps what you're doing today has changed, but I believe that little oasis in your heart is still there.

Why must entrepreneurs hold onto their original intention? Because trends are always changing. Without that anchor, it's easy to get pulled in every direction — whether by investment trends, startup trends, or profit-driven fads. If you have a clear mission in your heart, you can ride the tailwinds that trends bring to move forward. With original intention, you stay much steadier.

What our team believes is that the endpoint of technology should be making people more relaxed, not turning them into more efficient "super workhorses."

The biggest bottleneck and dilemma for many technically-minded founders is the urge to show off their "muscles" — I've trained for years, I need to demonstrate it. But the core of technology is solving problems. If it doesn't solve a problem, it belongs back in the lab to incubate further.

Technology isn't for showing off, it's for solving problems — and this is especially true for hardware. Everyone's original intention may differ, but one thing is universal: we increasingly need to follow first principles. We must deliver value, must output things that genuinely solve problems, bring joy, or provide solutions.

Flashiness may attract some people, but the core is always product strength. In this era, a first-generation product might take off quickly through some fortuitous moment — going viral on TikTok or Kickstarter; but the second and third generation products are the real test of whether a hardware company's founding team and core capabilities have what it takes.

Original intention acts as a filter, helping you strip away the noise and focus on what truly matters. Like sailing across a vast ocean, it keeps you steady and moving forward, however large the waves.

In this era of rapid forward motion, many people lack patience because they want to see results quickly, unwilling to wait and invest too much. But hardware is actually a compounding industry — the deeper you build, the more resilient you become, which requires founding teams to maintain ample patience. The core of patience is deep digging, going deep.

If you come from a technical background, there are several things you need to do thoroughly:

First, intensively focus on quality. For hardware, quality is paramount — you don't get many chances to fail, the market won't give you multiple opportunities to refute user feedback, and production lines can't be adjusted back and forth.

I know for many entrepreneurs, the dilemma is whether to "seize the window and not wait for perfection." Our answer is: you must seize the window, but you must also calm down and learn. Perfection means 100 points, 99 points; you can launch at 80 points to validate and iterate, but don't jump in at 30 points. In the past we've seen many products launch at 60 points — this isn't "not waiting for perfection," it's setting off without truly being ready.

From last November through the first half of this year, many people were talking about the "hundred glasses war," but count again now — how many eyewear products have actually established themselves in the market? The hardware industry has a distinctive characteristic: as long as you're alive, there's opportunity. The prerequisite is surviving first.

Here's a small example: in 2019, I brought products back to China to pivot, right when the company was at a low point compounded by the pandemic outbreak, on the brink of life and death. Seeing the situation in Singapore and other overseas regions, we made a decision to build a temperature-checking glasses. Because our product's foundational form was solid, we were able to add the functional module very quickly. You could say these glasses won the company precious windows for survival and development.

So, when opportunity comes, can you only seize it if your product is ready?

Second, engineering depth — this is where hardware moats begin. There are many product forms on the market, but differentiation often comes down to engineering depth. Engineering is an extremely complex, structured discipline requiring long-term thinking; it depends on the combination of hardware, components, algorithms, and all other links.

Over the past decade, Rokid has accumulated roughly 1,000+ intellectual property rights, including 547 patents domestically and internationally, and I was fortunate to participate in over 30 of them. These patents concentrate on human-computer interaction, optics, acoustics, spatial computing, and OS capabilities. Patents aren't just for protecting yourself — they're for protecting yourself from being taken down by others.

Patience also means focus and convergence. Many technically-minded people easily feel they can do everything, especially in the AI era. But what you're good at may amount to nothing until validated and feedbacked by the market. Like not having solved the problem or written on the blackboard yet — no one can prove your ability is real. So you must jump into the market to validate. Only in real storms do you know which way the wind blows.

So how do you find what to focus on when everyone is so busy? Use two key dimensions to judge: time and resources. We now operate on a "211" work model: two iterations per month, iterating every two weeks — one week to validate, one week to review; rapidly responding to user feedback, quickly knowing what to do next; then continuing to iterate. Glasses are extremely precise instruments. For software or smaller products, iteration cycles can be shorter.

Hardware is a compounding industry — the deeper you build, the more resilient you become. Look at DJI, Insta360 — though neither has been around as long as tech giants, they've iterated tremendously, with many deeply built layers, requiring great patience to achieve.

Also, particularly important is "going global," but its core isn't simply saying "I want to expand overseas," nor is it just marketing on TikTok. It's about "embracing ecosystems." A hardware company can't reach every place itself; if you only work with carriers and distributors, it's hard to build a good ecosystem. You need to embrace all possibilities for connection, even giving partners high percentages of profits and margins to help them push and complete the entire ecosystem. But the prerequisite is strong product capability — if product capability is weak, this is hard to drive.

Many friends often ask me: the big players are all coming in, what should we startups do? Every hardware manufacturer probably faces this question.

First, small-scale revenue isn't what major companies focus on. The key indicator of whether a product has truly entered competitive phase is whether it has achieved mass production; as long as it remains in the lab or validation stage, the market landscape isn't set, and opportunity still exists.

Second, look at ecosystems. Major players push products through their own platforms and can't accommodate too many third-party APIs. Our products have Doubao, Moonshot AI, Zhipu, StepFun, and all AI companies' applications, and we'll also have GPT, Google, and all AI companies embracing us. Each third-party API has its own specialized strengths. The eternal core for startups is this — you're not independent of the entire ecosystem; you can embrace everyone to create and push your own ecosystem.

Whereas major players push their own products based on their own platforms, so this is also a survival principle for crossing cycles. For example, when B-side clients choose a product, they want to select something less invasive or monopolistic.

Of course, I haven't mentioned whether luck plays a role here — luck is what all entrepreneurs need most. Everyone's a good student, so why are you an 80 and they're a 90? Maybe just because they guessed the exam questions right. A subject can't be determined by a two-hour exam to decide your fate and ability. But that's the reality — when a trend hits, you have only a week, a month to get your brand directly into the market. Maybe you take off, maybe you die.

I've been an entrepreneur for nearly 20 years, and I've always enjoyed this feeling of "dancing between life and death" — probably driven by dopamine and endorphins.

Also, how do small factories compete with big ones? When I did poorly on exams as a child, my mother would ask me one thing: "Subjective reason or objective reason — is this your problem or someone else's?" Many of us would say it's someone else's problem: the pencil wasn't good, the book wasn't clear, the teacher didn't explain well. But ultimately it all comes down to your own problem. Entrepreneurs have no excuses — only your own problems, or your team's problems, or you simply shouldn't be an entrepreneur, you simply haven't reached that moment.

You need judgment, not to treat major players as some flood or monster when they arrive. In our view, major players often help startups push a wave of attention, but opportunity is left open for everyone to compete for. So maintain hunger, and confront your own human nature. Whether to be more greedy or more laid-back — there's no right or wrong answer; your choices are what make the entrepreneurial environment diverse. I rise, you fall; I rise again, you fall again — this is what drives industry and technology forward.

There's a major structural opportunity in the industry right now: AI needs hardware carriers. When AI only exists in PCs or phones, its imagination isn't truly unlocked — people regularly use fewer than 50 apps, but there are millions in the app store.

We need new carriers, an independent private hardware matrix undisturbed by work or other information. This is the future of AI hardware, and also the backdrop for the rise of fields like health AI, companion AI toys, and future home robots.

Of course, beyond serving the mass market, those overlooked niche groups also contain enormous opportunity. We have a nonprofit project called "Bright Eyes," developing glasses for visually impaired people that can real-time identify people, objects, and scenes ahead of them.**

For us, AR glasses' photo recognition capability might just be a 15%–20% product premium; but for the visually impaired, it directly determines whether they can access environmental information, whether they can "see" — the value delivered is an order-of-magnitude change, a leap from zero to one. Where there's real value, real opportunity is born.

So everyone should embrace multimodal and wearable technology, and also pay attention to specific niche groups. For example, traditional white canes only know there's an object ahead, but can't identify what it specifically is. What we do isn't just giving blind people convenience in life, but also giving them more possibility to create value. For instance, they can share their content on platforms and earn some income.

When hardware reaches a certain point, it's no longer just about yourself. Now all leading hardware companies are opening their ecosystems outward, letting others grow, earn money, and achieve commercialization on their platforms **— this is also what we're doing.

I want to close by sharing two phrases: one is **Stay Human — I hope everyone maintains the human part of us as carbon-based life, this is extremely important motivation;

The other is our company slogan — Leave Nobody Behind, which reflects our original intention. Everything we've done is fundamentally about letting technology serve everyone and letting everyone enjoy technology.

There are two key points here: one is "everyone" — not just current mainstream users, but also all the overlooked groups I just mentioned; the other is "enjoy," which is actually very difficult. For a tech product to make people feel enjoyment means its usability, friendliness, and price must all be reasonable, and it must continuously deliver value.

We believe that no industry ever advances through one company alone — it must rely on many participants moving forward together. I hope all entrepreneurs can look back at their original intention, maintain patience, and give themselves and the industry more confidence.

---