The AI hardware sector has never been so crowded.
From the conversational smart ring Stream created by two former Meta employees to glasses integrated with AI assistants and portable recording cards, even pendants and wristbands, tech companies are attempting to embed artificial intelligence into all wearable devices. Even if Musk does not make the assertion that “edge intelligence will rise,” a hardware competition around interactive entry points has already fully ignited in Silicon Valley.
Domestic giants are not to be outdone. At the end of August, during a major version release codenamed “Fern,” Dingtalk’s first AI hardware product, the Dingtalk A1, was unveiled, carrying Dingtalk’s hardware genes. Internally at Alibaba, this is a team that has won battles in hardware. Externally, many believe this is merely a strike in the smart hardware sector, and that the “big companies entering the field” only adds more attention to an already crowded market.
But why is Dingtalk entering the AI hardware space at this time? Many may have misunderstood.
While many are raising grand narratives about how AI will change the world, which attracts attention and money, Dingtalk is somewhat different.
If AI can change the world, it must involve the participation of the majority, but first, people must help AI understand the physical world. To enable everyone to actively use tools to help AI enhance its understanding of the physical world, universal access to tools is essential.
In Dingtalk’s view, so-called AI universality cannot be achieved through top-down indoctrination by a few companies. It is shaped by the experiences of every ordinary employee: it may be the construction of a form, a point-to-point response from Dingtalk customer service, or a Dingtalk programmer visiting clients across various locations. All occurrences are trivial and quiet, like a gentle stream.
However, if one can understand Dingtalk’s long-standing commitment more deeply, they may agree on this point: the Dingtalk A1 is not born to defeat a specific company or to be a blockbuster product; it is the first offspring of Dingtalk’s AI universality concept and the starting point of a long journey.
“The Silent Majority” and the Ever-Present AI
By the end of 2024, there will be over 60 million enterprises in the country, with small and medium-sized enterprises accounting for the vast majority of all enterprises, sustaining the vitality of the Chinese economy.
However, when the AI wave roars in, some small and micro enterprises are far from the sound of the tide. They fear being left behind by the times; “we must do AI” has become a consensus, but how to do it and how to use it remains a mystery. They have become the “silent majority” in the current AI frenzy.
Here, the first barrier is the technical threshold.
Currently, AI large models are still productivity tools for a select few; they require constant tuning and optimization. For a small company that lacks personnel, funds, and time, the application threshold is evidently very high. For example, in Yiwu, enterprises with fewer than ten employees account for the vast majority; they do not have a clear “management department,” rarely hold meetings, and cannot even talk about processes.
“We see AI dominating the world every day on short video apps, but when you actually enter a company, you find that many companies can’t even use basic office software properly,” observed a Dingtalk service provider.
Many enterprises have no idea how to implement AI large models and lack the extra energy to adapt to various AI tools. These users expect a product that can solve daily problems in the simplest and most efficient way, visibly and tangibly improving productivity.
This is the most authentic enterprise-level market in China, rich in layers and complex in environment.
Therefore, for AI to truly become universal, it is not about making the models more powerful, but about reducing the threshold to zero.
As a product that has been entrenched in this market for many years, Dingtalk is very clear at this moment that software as an entry point has already reached its limits: software has a very high startup cost, just as one customer said during Dingtalk’s team research: “When I really want to record something, I simply don’t have time to open the software.”
This is also the origin of the Dingtalk A1, whose mission is clear: to ensure that “AI is always present.”
Just as Edison lit the first incandescent lamp of the electrical era, and Apple replaced keyboards with touch on the iPhone to usher in the mobile internet era, every leap in technological revolution ultimately needs to be carried by physical hardware. After all, humans can only interact and connect more densely with tangible objects.
As Dingtalk’s first AI hardware, the Dingtalk A1 is presented in the form of a card-style recording pen. This is mainly based on the consideration of “non-perception” in use. AI hardware is a new thing, and the interaction interface must be simple and clear enough to be quickly grasped. The Dingtalk A1 fits with mobile phones, achieving all-scenario, all-weather portability, and can be operated with one button; users only need to do two things: press the button and start.
For example, a sales company manager might think of tasks or solutions and casually open the A1 to record. “It’s much more convenient than opening my phone to find an app; no one will find it strange, and it’s convenient for me too.”
If we look back a few years from now, the significance of the Dingtalk A1 goes far beyond just being a card recording pen; it is more like a bridge connecting the data world and the physical world, and a key position in the era of “spatial intelligence.”
The First Key to Unlocking Spatial Intelligence
On November 10, Professor Fei-Fei Li from Stanford University published an article proposing the concept of “spatial intelligence.” She believes this will become the next peak of AI technology. Currently, AI systems represented by large language models, while proficient in generating text and images, still remain in the “world of language” and lack a true understanding of real space, physical laws, and causal relationships.
The interactions between people and between people and objects constitute a complexity that far exceeds language. Undoubtedly, the new generation of AI large models must possess agency, perceiving, reasoning, and acting in the real world like humans.
Currently, we enrich input methods through cameras, microphones, sensors, etc. Through voice intelligence, visual intelligence, and tactile intelligence, AI gradually opens its “five senses,” step by step understanding the meaning of human behavior in physical space, ultimately “reconstructing” the world on geometric and physical levels.
If we agree with this path, the Dingtalk A1 is undoubtedly the beginning of establishing voice intelligence.
Through this barrier-free interface, AI can obtain uninterrupted, unstructured, multi-dimensional spatial information.
Compared to ordinary smart devices that can only access the microphone, the Dingtalk A1 is equipped with five omnidirectional microphones and one bone conduction microphone, capable of recognizing sounds within an 8-meter range. The result is that the breadth and depth of information acquisition far exceed what a mobile phone can achieve.
If previously the data injected into enterprise-level large models was a trickle, with the Dingtalk A1, the data will flow like a mighty river.
For example, the AI product “AI Listening and Recording” that works with the Dingtalk A1 has added a “visual recording” feature, utilizing the five microphones of the Dingtalk A1 to recognize different speakers by voiceprint and their positions in space. When reviewing recordings, the interface will visually present who spoke when and where, restoring the meeting scene.
Ultimately, Dingtalk hopes that this information can be distilled into “knowledge” and “wisdom,” improving the productivity of ordinary employees and business owners in work scenarios, helping enterprises build their own AI capabilities, forming a complete closed loop from data aggregation to model building, assisted decision-making, and feedback learning.
Clearly, in the formation of this closed loop, the relationship between humans and AI has evolved from a one-way input-output to a two-way human-machine collaboration. With Dingtalk achieving hardware-software integration, Dingtalk AI can process the information collected by Dingtalk and embed it into the entire workflow.
For Dingtalk’s users, assisting in “decision-making and action” is the most visible change, transforming Dingtalk from a passive work software into a self-driven advisor and consultant.
One case is a startup entrepreneur in the RV import-export business who called overseas clients to take notes and translate, then wrote emails himself. Now, after meetings using the Dingtalk A1, phone conferences are directly converted to Chinese in real-time, and emails are generated based on meeting requirements; all he has to do is make final edits and approvals.
For employees, this efficiency boost does not incur additional time costs and helps them continuously refine their business skills. For instance, a social worker usually records meetings with the Dingtalk A1, and Dingtalk can organize the text through AI, providing analysis and summaries based on meeting minutes. For example, the AI assistant will inform him/her of the methods used in completing tasks, what aspects have been done well, and what areas still have shortcomings.
This is a silent revolution led by Dingtalk, liberating the productivity of managers and frontline employees in small and medium-sized enterprises with AI in countless offices and warehouses. As a Dingtalk service center staff member said: “What Dingtalk is doing is essentially laying down pipelines. Only by ensuring these pipelines are clear and providing excellent service can water, electricity, and coal—namely AI capabilities, computing power, and data—truly flow into every small enterprise.”
The First AI Hardware Truly Released by Alibaba
In the enterprise-level AI market, the source of data comes from tens of millions of enterprises across the country; data is undoubtedly the raw material for AI applications, while computing power is the foundation of the AI world.
Globally, the arms race for AI computing power and talent has reached its midpoint. Entering the era of spatial intelligence AI, it is foreseeable that the territories of major forces will continue to expand.
Taking enterprise-level AI hardware as an example, China’s Pearl River Delta’s strong industrial chain has completely dissolved the barriers to hardware manufacturing. For just a few dozen yuan, a card-style recording pen can be produced in Shenzhen. Clearly, the ultimate competition in AI hardware still lies in AI software capabilities, or even the development capabilities of the entire system and ecosystem.
In the end, only the continuously advancing strong will remain in the field. For this reason, the responsibility for AI universality must be undertaken by national-level applications like Dingtalk. After all, Dingtalk’s AI is backed by the computing power, technology, and talent of the entire Alibaba Group.
As an important layout for Alibaba, Dingtalk most intuitively showcases this strength. At this year’s launch conference, Dingtalk released over ten AI products, including Dingtalk One, AI Inquiry, AI Forms, AI Listening and Recording, and the smart hardware Dingtalk A1. This quantity, quality, and speed of product matrix release is something that startup companies cannot reach.
Backed by Alibaba’s technology, Dingtalk’s AI products exhibit astonishing efficiency. The Dingtalk AI Forms and Alibaba Cloud ADB-PG database team collaborated to launch the integrated storage and computing architecture O-Table, which supports real-time updates of millions of rows in a single table, requiring only seconds.
Strictly speaking, the Dingtalk A1 is also the first AI hardware truly released by Alibaba. It is also the first card-style recording pen in the industry to incorporate real-time transcription functionality. Real-time means there is no time for the AI model to modify, resulting in a very low error rate. On the computing power side, if post-transcription is changed to synchronous transcription, the model must possess the computing power to handle high-concurrency scenarios.
Currently, market card-style recording pens can only upload recordings to the cloud after recording, and then process them through external large models. The reason they do not include real-time transcription is that this functionality would exponentially increase the difficulty and cost of transcription.
Moreover, card recorders are generally billed by duration, only calculating the final uploaded recording duration to the cloud. Once real-time transcription is added, the entire billing model needs to be restructured. Therefore, the current lack of this seemingly simple function is actually the inevitable result of the computing power gap.
The responsibility of a pioneer has fallen on Dingtalk, as it is precisely these barriers that allow the Dingtalk A1 to dare to attempt real-time transcription. Furthermore, the Dingtalk A1 offers 1000 minutes of free usage time, the highest in the industry, and its real-time transcription does not count against this time.
Undoubtedly, the Dingtalk A1, as a hardware product, is not a high ROI product. The computing power costs behind the product are difficult to dilute in the short term.
In fact, Dingtalk has also pioneered the use of an AI pay-per-performance model in the software sector. At a time when enterprises are still hesitant about paid AI products, as a leading enterprise, Dingtalk is willing to cover the “entrance fee” for small companies, with the ultimate goal of enabling enterprises to “dare to use, be able to use, and afford AI.”
This can be seen as a certain responsibility that Dingtalk has taken on for AI universality.
“They really want to make this thing work well”
“Many people believe that elites can change the world; that is not the case; it is the grounded elites who change the world.” After the Dingtalk A1 launch in August, Dingtalk founder Wu Zhao said in an interview with 36Kr.
Political philosopher Michael Sandel has also proposed a viewpoint that elites succeed not only through their own efforts but also due to factors such as background, luck, and timing. Once elites take this luck for granted, they fall into “elite arrogance.”
As one of Alibaba’s core products, Dingtalk can naturally mobilize a large number of resources and funds; its starting line is already far ahead of its peers.
For example, the more than ten products released by Dingtalk this time are all supported by Alibaba’s grand backing. Just taking the “AI Listening and Recording” product as an example, Dingtalk has completed over 100 million hours of audio and video data training with the Tongyi Laboratory. The final result is an accuracy rate of 90% for recognizing over 30 dialects in China and 140 languages worldwide—they completed in one month what would take a startup several years. This product capability, when applied to the Dingtalk A1 hardware, can quickly create a gap in voice transcription capabilities compared to others.
This is a reality that is easy to be proud of: the success of a product certainly involves the team’s own efforts, but one cannot deny Dingtalk’s “inherent superiority” in resource endowment.
However, the Dingtalk team always maintains a distance from this “sense of superiority,” striving to uphold a humble and grounded style.
If you delve into Dingtalk’s internal workings, you will find a team that is very “grounded” from top management to grassroots employees. Their management and R&D personnel must personally visit clients, listening to many seemingly trivial questions from customers, including how to find the entry point, why there are charges, etc. The product team for the Dingtalk A1 is no exception; users have praised this team on Xiaohongshu, noting that official personnel reply to user questions one by one in the community every day, showing that they really want to make this product work well.
In the recently concluded Double 11 shopping festival, the Dingtalk A1 has become a dark horse in the recording pen category, achieving leading sales and transaction volume on platforms like Douyin and Tmall, with over ten thousand units sold in just one week, leaving a batch of new and old players behind. In just three months since its launch, Dingtalk has begun to transform this sector with a remarkable comeback.
For Dingtalk, the Dingtalk A1 is just the beginning of truly defining a new way of working in the AI era, and also the beginning of constructing a piece of the puzzle for AI to change human life. Dingtalk’s hundreds of millions of users, and even competitors, are all part of this group of builders; they will compete, collaborate, but all will head towards a common future.