The OS Upgrade That Quietly Redrew the Rules of Business

I usually write about business, and I rarely make a point of following a specific operating‑system update.

The reason is simple: for most people, a system upgrade means a new look, a few extra features — no big deal. Even for developers, the inevitable arrives: more APIs, updated tools — also no big deal.

That judgment held for a long time. Until I went to the HDD·HarmonyOS Innovation Forum in Chengdu.

My reason for going was casual — a friend invited me, and it was only 800 meters from where I live. I walked in with the mindset of “since it’s that close, might as well listen.” And then I realised my previous judgment was probably wrong.

The core of this HarmonyOS upgrade isn’t how much the UI changed, nor how many new interfaces were added.

The core is three things: how services are connected, how they are invoked, and how they are delivered. The system‑level AI upgrade is the foundation; on top of that, these three aspects have been reorganised.

And when those three things move, business rules move with them.

The entry point has changed. The way products are built, distributed, and competed over will all change accordingly.

I’ll start with a few technical changes from the forum and talk about the undercurrents they might trigger in business.

## 02 Star River Interconnection: Devices go from “each on its own” to “one network”

Start with the first change: how devices collaborate.

An operating system’s core job is essentially threefold — manage hardware, run applications, provide system capabilities. That’s the textbook definition, unchanged for decades.

But nowadays, everyone has more and more devices. Phones, computers, tablets, watches, glasses, cars: one person can easily own three to five, maybe ten. And then comes the trouble.

How do devices coordinate with each other? How do applications pass capabilities between them? When a user casually says something, how does the system know which service to call?

Previously, all these were the developer’s problem. You want cross‑device functionality? Fine — handle device discovery, network connection setup, connection state management, reconnection on failure, data transfer, permission validation, format conversion, and so on. By the time you’ve dealt with all that, you’ve spent most of your energy on infrastructure before even writing a line of business logic.

HarmonyOS 7 makes a key move: it takes back part of that complexity from developers and puts it back into the system layer.

The upgraded Star River Interconnection architecture supports smooth cross‑system, cross‑device, cross‑scenario connectivity. These terms sound technical; translated into what users actually feel: file transfers without repeated pairing, device switching without restarting, cross‑screen collaboration that no longer feels like operating several machines that don’t recognise each other.

But for the business world, the value is not in the smoothness of user experience.

The value lies in this: the developer’s unit of design has changed.

Previously, the design unit was “a single app on a single device.” Now that the system takes over the messy work of connection and scheduling, the design unit can expand to “a complete scenario.”

What does that mean? Your product is no longer confined to a phone screen. It can simultaneously orchestrate phones, tablets, car consoles, and watches to accomplish one thing together. Your competitor is no longer another similar app, but another solution that better fulfills a user’s scenario — even if it uses three completely different devices.

The expansion of the design unit means the boundaries of competition are being redrawn.

03 Huawei Share: The imagination around content flow goes far beyond “file transfer”

Once devices are connected, the second question is: what to transfer and how.

Huawei Share’s cross‑protocol capability deserves to be singled out. Its relationship with Star River Interconnection works like this: Star River Interconnection handles how devices “shake hands,” and Huawei Share handles what is handed over and how after the handshake.

Common problems have been absorbed by the system. Data format conversion, protocol compatibility, resumable transfers — all previously handled by developers — are now packaged and ready to use.

The imagination this brings to business scenarios is far greater than just “file sharing is a bit faster.”

Here are a few scenarios.

Gaming scenario. Your friend has updated the game, but you haven’t downloaded it yet. Previously, you’d either wait for him to send the installation package — Bluetooth too slow, cloud storage too roundabout, chat apps limited by file size — or you’d download it yourself, wasting tens of minutes. Now? One tap, and the resource pack comes straight over.

That’s just the basics. Go further: you discover a hidden route in an open‑world game. One tap, and what your friend gets isn’t a few map screenshots, but a complete exploration route including coordinates, markers along the way, and mission order. The two of you can even merge the areas each has discovered, forming a jointly explored map. The interactivity is maximised.

Travel scenario. Two people go on a trip; one has already planned a three‑day itinerary. One tap, and the other person doesn’t get a travel guide to read, but an executable plan: hotel addresses automatically marked on maps, train times automatically added to calendars, restaurants saved to favourites, attraction tickets available in the wallet. It transforms from “reading information” to “gaining capability.”

Membership scenario. This one is especially instructive.

The most painful problem for membership systems: how to spread via social sharing without breaking the payment model?

Previously, even sharing via a link involved cumbersome steps — having the intention to share, opening the app to find the entry, generating a link, switching to WeChat, sending, waiting for the other to open it. Each step caused drop‑off.

Now: have the intention to share, one tap, done.

Even better, the other person can fully experience the membership benefits without actually being a member. That’s a natural hook. Once they’ve experienced it, the conversion probability is naturally higher.

On the technical level, it’s just about solving a permission‑boundary issue. But on the business level, the acquisition chain has been compressed to a single physical touch.

04 Xiaoyi: When the entry point changes from “you find it” to “it finds you”

The two changes above address how devices connect and how content flows.

One step further, there’s a more fundamental question: how are users’ needs actually understood and fulfilled?

Xiaoyi’s Agent and Skill are designed to solve exactly that.

The user speaks naturally about what they want to do. Xiaoyi’s Agent understands intent, breaks down tasks, and plans paths. Skills provide specific capabilities that can be invoked. The foundation is laid by developers through ArkTS and various Kits.

Behind this is a major trend: from the App era to the “say‑it‑and‑it‑happens” era.

Many products now claim to be “AI native,” but most still stop at “adding a chat box inside the app.” Xiaoyi’s logic is different — it combines the intent‑understanding capability of large models with workflow orchestration, breaking complex requests into multiple steps and invoking corresponding services to complete them.

More crucially, it can combine system‑level perception data, context, and cross‑device state to judge what the user is more likely to need at that moment. In other words, it doesn’t mechanically execute instructions; it operates with understanding.

The logic of the entry point has changed. Previously, you looked for services — unlock the phone, find the icon, open the app, locate the function. Now, services look for you — you say something, the system understands and schedules the capability over.

It also means something else: cross‑Agent invocation (A2A).

Xiaoyi and Damai’s Agent connect via cloud‑side A2A to cover the full chain of Q&A, ordering, and payment. The strategic choice here is telling: not building all industry Skills internally, but defining interface standards and letting industry Agents plug in themselves.

AI is already challenging the concept of apps. Agents may become the new super‑entry point. HarmonyOS 7’s layout is preparing for this major trend.

05 ArkTS: Not a technical issue, but a business‑implementation issue

We’ve talked about many scenarios and imaginative possibilities, but if the development barrier is too high, everything remains a castle in the air.

ArkTS and the supporting development tools determine one thing: whether these capabilities can actually make it into code.

Huawei clearly understands this. The latest AI‑assisted development tools allow developers to complete testing by introducing just 3 Kits and writing 5 lines of code. The newly released HarmonyOS Intelligent Agent Framework 2.0 (HMAF 2.0) achieves over a 90% success rate on complex tasks.

The low barrier has led to rapid ecosystem growth. In just half a year, HarmonyOS applications and services grew from 350,000 to 400,000. Some apps are already exploring the upgrade from “manually opening an app” to “accomplishing a task with a single sentence.”

The validation of commercial returns is even more telling.

A developer shared on the forum: a 7‑person team called “Wonderful Toolbox” achieved over 500,000 yuan in monthly revenue within the HarmonyOS ecosystem, with a conversion rate far higher than on Android and iOS. Apps like “Jianxun” are also deeply integrating with AI Agents to explore the experience upgrade from “manually launching an app” to “completing a task in one sentence.”

This is not an isolated case; it’s a signal. It shows that this ecosystem already has the ability to convert traffic into revenue.

05 A judgment that needs revising

At this point, I went back and flipped through another article I wrote two years ago about operating systems. The core judgment of that article was: changes at the operating‑system level take at least three to five years to show business impact, and early entrants are mostly “betting on a possibility.”

That judgment was correct for the early days of Android and iOS. But the data from HarmonyOS 7 made me rethink this.

“Wonderful Toolbox” — 7 people, 500,000 yuan monthly revenue, conversion rates crushing Android and iOS. Time span? Half a year.

Half a year.

Not three to five years. Half a year.

The implication: the “wait until the ecosystem matures before entering” strategy might incur opportunity costs in this cycle. By the time you see others making money and join, you may have already passed the steepest part of the traffic dividend.

This judgment revision may be more important than any single technical change — it’s not saying “HarmonyOS is great,” but rather “the shape of the window has changed.”

06 The shift: the rules of competition have changed

Putting these pieces together, the logic of HarmonyOS 7’s iteration becomes clear:

Star River Interconnection makes devices collaborate; Huawei Share makes content and capabilities flow; Xiaoyi enables the system to understand what users want; ArkTS lets developers bring ideas to life.

These four things combined change the role of the app — it is no longer an icon waiting for the user to tap, but a service capability that the system can understand, invoke, and orchestrate.

This shift points to one conclusion: the rules of competition are being rewritten.

Previously, competition happened on app store rankings. Whoever had more downloads, better retention, and higher monetisation won.

Now, the entry point has become system‑level intent. Users no longer say “I want to open App X,” but “I want to do task Y.” The system understands and invokes the most suitable capability to fulfil it.

As a result, competition moves from “rankings in the app store” to a multi‑dimensional contest involving device connectivity, scenario coordination, and intelligent service invocation.

Which dimensions do you have an advantage in? Can your capability be discovered and invoked by the system? Can your service flow smoothly across multiple devices? These questions may matter more than “how good the app looks.”

07 Data points to track

Finally, here are a few data points. They are not conclusions, but they are useful for anyone making business decisions:

HarmonyOS 6 device installations — 66 million

Total devices in the HarmonyOS ecosystem — 1.3 billion

Registered developers — 11 million

Applications and services — over 400,000, up 50,000 in half a year

Average daily app downloads — over 200 million

Xiaoyi daily active users — 180 million, with 3 billion daily invocations

What these numbers signify: the HarmonyOS ecosystem has moved from a “shall we try it” observation phase to a “needs serious planning” strategic phase.

Some developers have validated commercial returns, data that proves traffic scale, and technical pathways that lower the entry barrier. The remaining question is: who will figure out first how to integrate their business capabilities into this new system.

Conclusion

Most of the time, an operating‑system upgrade really isn’t worth a dedicated article.

But this time is different.

HarmonyOS 7 is redefining how exactly a service should reach its users.

The answer to that question is shifting from “let users find your app” to “let the system understand user needs and then call your capability.”

The entry point has changed. The rules have changed. The window won’t stay open forever.

That conference 800 meters from my place changed my judgment about operating‑system upgrades. I hope this article helps you rethink a few things as well.