At a time when AI hardware is constrained by energy consumption, privacy anxiety, transmission lag and global applicability, an industry rookie named "Ai Ling Technology" is breaking the game with its unique technology brand positioning and is determined to become the world's leading provider of end-side AI chips and scenario-based solutions.

Founder & CEO Zhu Yuhao-Director of Shenzhen Engineering Laboratory, Shenzhen High-level Talent, Doctor of Business Administration, once served as the head of research and development of domestic communication industry head enterprises (A-share listed companies), built the core research and development capability chassis in many years of practice, and also mastered the full link logic of products from technical blueprint to market landing.

Zhu Yuhao, Founder & CEO of Ailyworld Technology

This time, he jumped out of the comfort zone of the mature platform and took the lead in setting up a professional team to start a business. Instead of following the trend and chasing hot spots, he focused on the layout of four core technologies, pointing directly to the existing difficulties in the industry:

As a start-up company, Ailyworld Technology, with its end-side AI chip technology and scenario-based solution, stands out in the competition with many established smart hardware companies and wins the cooperation of a head mobile phone manufacturer.

Its latest AI portable assistant-Note Pro recording card is charged wirelessly with the world's thinnest body of 2.89mm, surpassing its peers in many product performances and setting a new benchmark in the industry.

As a new force in the AI industry, Ai Ling's position is not ToC but ToB. In Zhu Yuhao's view, ToB is no less than ToC, and Ailyworld has a wider customer base and more increments than Plaud. "Our benchmarking company is not Plaud. Elin wants a bigger market." Zhu Yuhao said.

As early as 10 years ago, Zhu Yuhao saw the first wave of smart hardware start-up opportunities, when Stone Technology, Yunzhisheng and other start-up companies have emerged one after another. "Ten years ago, this wave of intelligent hardware mainly brought connectivity to the hardware and had its own APP," Zhu Yuhao said. "But this wave of AI hardware has a greater chance, and artificial intelligence enabling hardware has a lot of imagination."

After ten years of sharpening a sword, when Zhu Yuhao led the team to officially enter the game, Stone Technology has long relied on its brilliant market performance to become the first individual stock with a share price of more than 1,000 yuan in 20 years, and was given the title of "sweeping the floor" by the capital market. The market value of AI solution provider Yunzhisheng, which focuses on the dual scenarios of "daily life medical health", has steadily climbed to more than 40 billion, showing strong market recognition.

How will the industry landscape evolve over the next decade? The answer to this question may be hidden in the trajectory of Ai Ling's technology.

The following is the content of the dialogue between Leifeng Net Whale Rhinoceros and Zhu Yuhao, CEO of Ailyworld Technology. Whale Rhinoceros has been edited and arranged without changing its original intention.

Whale rhinoceros: you choose to join the field of AI hardware entrepreneurship, how did the initial entrepreneurial idea germinate? Why choose this time node to start a business?

Zhu Yuhao:Ten years ago, there was a wave of smart hardware startups in the industry. Many of the founders of those smart hardware startups were friends they knew, and they knew their stories better. I also saw that they had made achievements and contributions to the industry. At that time, I planted the seeds of entrepreneurship in my heart.

This time I chose to start a business, yes.Right time, right place, with right people.。

Timing, referring to trends, I observe two major industrial opportunities: one is that traditional products are accelerating their upgrade to AI empowerment, and the other is the continued release of global market dividends under the wave of products going out to sea. There are a lot of opportunities in these two trends.

Land, refers to the technical precipitation. We have independently mastered a series of key technologies with differentiated advantages, which can directly attack the core pain points faced by the current industry in AI hardware applications.

People and means teams and resources. We 've put together a very good team. The founding team has deep accumulation in the fields of AI large model development and application, acoustics and natural language processing, chip custom design and development, low power consumption technology, wireless hardware products, etc. It has also deposited many upstream and downstream resources in AI audio and video industry, and can rapidly promote the transformation of technology into products.

Whale rhinoceros: What are the different ways to position ToB by Ai Ling Technology?

Zhu Yuhao:I have a deep understanding and practical accumulation of ToB mode. In the past, when I was a professional manager, I was in charge of the wireless application division and the new product development department successively. I had a lot of successful trading experience, including starting the new product business from 0 to billions of levels. Many products have a market share of Top1 or Top2 in the world and have gone through the whole process of business from 0 to 1 and from 1 to N.

The newly founded Ai Ling Technology also positions ToB, but has a different style of play--Less factories, more technology. In manufacturing, we choose to cooperate with various foundry factories. They are more professional. We can also have more flexible product strategies and supply chain choices, which are more suitable for the characteristics of AI hardware circuits.

We attach great importance to the construction of core technology, independently mastered the algorithm and chip, end-side model, low power consumption, full wireless and other underlying technologies, and then go up to form a full stack of technical solutions, we hope to establish a continuous differentiation advantage through the technical threshold.

Whale rhinoceros: We know that Ailyworld Technology has a clear layout in the AI direction on the end side. The name "Ailyworld" also sounds very recognizable. Can you share with us the original intention of brand naming?

Zhu Yuhao:Slogan of AI Ling Company is "AI Listening to the World/AI Listen To The World", which contains two-way value expression:

On the one hand, it expresses the technical hard power-not only represents the technical direction of our deep-rooted global language speech algorithms, but also clarifies the strategic positioning of the company's AI scenario-based solutions for the global market since Day1;

On the other hand, it conveys the soft power of service-we always adhere to the attitude of "listening first" and cherish the voice of every customer and user. This "listening" is the core logic of our "user-oriented" product design. It is a direct manifestation of the corporate tenet of "customer first.

The company name "Ai Ling" means "AI listens to the world" and is a concentration of our ideas. "AILYWORLD", as our technology brand trademark, has become the concrete carrier of this concept and technical strength.

In addition, there is another meaning, which stems from the tribute to the father of AI. The homophonic "Ai Ling" comes from Alan Turing, which not only expresses respect for the pioneer, but also carries the company's determination to continuously explore and pursue excellence in the field of artificial intelligence.

Let "Einstein" do scientific research, let "high school students" do housework

Whale rhinoceros: You mentioned that there are many pain points in the current AI hardware industry that have not been well solved. What are the specific points?

Zhu Yuhao:The integration of AI and the physical world holds unlimited potential, but the process of integrating AI into hardware still faces many challenges. Including efficiency, cost, experience, etc., as far as I can see, the AI hardware industry has the following four major pain points:

Global applicability, especially the demand gap for small languages.I have always believed that voice is the main entrance to interact with large models at present and for a long time in the future, and voice is the most convenient way to interact in many scenes.

China and the United States account for less than 22% of the global population, and many other regions are dominated by various minority languages. The most widely spoken population is Spanish, which exceeds 0.6 billion. Others like Arabic, Indonesian, Russian, Portuguese, etc. have a small population of hundreds of millions.

The penetration rate of AI voice in these areas is very low and is in the initial stage of intelligence. However, the core market of overseas large factories is in Europe and the United States, domestic manufacturers are limited to go to sea, and the investment in new development is huge, resulting in a very large demand gap for small languages, which has not been well satisfied.

There are many problems behind the big model.First of all, each large model has different capabilities, and how to select it is a difficult problem. Secondly, the payment mode is not smooth, AI products need continuous service, and backward payment needs to adapt to local payment methods and tax compliance. Third, we have to face users' concerns about data privacy and compliance review pressure from Europe, America and other countries. Fourth, the interaction cost of large models is relatively high, and high-frequency interaction requires additional subscriptions, the cost is more difficult for C- end users to accept.

Energy consumption and efficacy.The size of the wearable device is strictly limited, resulting in extremely low battery capacity, but the end-side AI task requires continuous computing support. In addition, the waste of computing power of AI hardware is common, such as voice wake-up, the number of wake-up times per day is relatively small, and a large amount of time is empty.

Selection of the device interface.The mainstream USB interface will increase the volume and weight. Frequent plugging and unplugging easily lead to interface damage. USB transmission of wearing equipment is slow, and there is also the anxiety that data lines need to be carried with them. The bottleneck of Bluetooth transmission speed and poor user experience; The integration of WiFi and 5G also has problems such as design complexity, global compatibility and certification.

Whale rhinoceros: Summarizes the current pain points in the industry, so how does Ailyworld Technology solve it?

Zhu Yuhao:Ai Ling Technology chooses to enter the game, we must grasp the pain points of the current industry, and make enough differentiation advantages in this, in order to have a chance of success.

global language speech algorithm:The core members of our team have more than 20 years of experience in overseas markets, have been deeply involved in overseas IoT projects, and have established in-depth cooperation with many international voice technology and data companies, laying the foundation for our rich overseas corpus resources.

We targeted developmentOffline Speech Recognition Algorithm for 36 Small LanguagesIt should be the most complete offline language in the industry, covering most countries with a large global population and a high degree of electronic intelligence, hoping to better solve the problem of global applicability.

Large vertical model:In view of the many challenges of large model applications, we first support the existing large cloud model platforms, including domestic bean bags/Tongyi/Yuanbao/Deepseek, overseas ChatGPT/Gemini, etc., and then continuously enrich the knowledge base and strengthen training for specific groups and vertical applications. At the same time, we will continue to accumulate user application data. Our vertical model has become a super assistant with higher and higher IQ. Different from the general big model in the cloud, it will know more about the vertical field and ourselves.

We will also make some AI capabilities that determine the scene to the end side. First, the data will not be in the cloud and will be processed directly locally to avoid users' privacy concerns. Second, we will save AI interaction costs and computing resources.The big cloud model has a high IQ brain, but some things don't need Einstein, only high school students can do it.

Adhere to the first principle, adhere to the pursuit of the ultimate high performance is our design philosophy.Our vertical large model reorganizes the end-cloud collaboration architecture to better enhance the user experience, helping large model vendors solve the problem of scenario landing and forming a continuous cooperative relationship with them.

Low Power Technology:The traditional CPU/DSP or general-purpose NPU architecture running end-side model is a large horse-drawn car with low energy efficiency ratio, while under the storage and calculation separation architecture, the power consumption of each calculation data between the memory and the calculation unit is several times that of the calculation itself, and the waste of invalid power consumption is prominent.

Ai Ling adopts an innovative chip architecture of dedicated low-power NPU and storage and calculation, combined with lightweight end-side AI algorithm and heterogeneous computing collaboration technology, so that 500mAh lithium battery can provide up to one month of voice wake-up battery life,Recording power consumption can be 1/4 to the industry average.

Full wireless design:Self-developed wireless fast connection wireless fast transmission wireless charging technology, high-speed WiFi is 50 times faster than Bluetooth transmission, faster than USB, away from cable troubles. Open up the big data traffic channel for AI hardware to interact with large models, greatly improving the AI application experience. For example, weThe AI recording card can upload 1 hour of recording data in 10 seconds, which is several times faster than the benchmark competition.

Whale rhinoceros: Does Ai Ling currently have a target company?

Zhu Yuhao:There are no very similar benchmarking companies. However, Yunzhisheng's business scope is more similar to ours. It is a company we are studying at the current stage. It has recently been listed in Hong Kong stocks and is currently valued at more than 40 billion, which has been recognized by the capital market. We will refer to the development path of Yunzhisheng, but hope that the financial development will go more steadily. In addition to Yunzhisheng, HKUST Xunfei is also the object of our study.

We should be a start-up company that rarely proposes to make a profit in the first year, choose the right and potential track in the direction, and maintain a sound financial development.

Small language offline voice to the Chinese manufacturing installed "world ear 」

Whale rhinoceros: your company mainly does small language offline speech recognition algorithm, compared to online speech recognition, what are the advantages?

Zhu Yuhao:The advantage of offline is that users do not need to connect to the distribution network, and the product cost is lower. here, we should first talk about an important phenomenon,Graffiti Smart has more than one billion wireless modules shipped, but the online opening rate is very low.With so much intelligent connection hardware on the market, the proportion of users who are really configured to connect to the Internet is very small. Most users do not have an IT background, and a slightly more complicated configuration will stop them, not to mention the stability of wireless connections, which will bring more confusion about commercial landing.

Based on this background, we have focused on strengthening the offline speech recognition algorithm, focusing on the blue ocean market of overseas small language countries, which are in the early stage of electronic to intelligent development. We have trained a standard entry database for a series of mainstream home appliance categories, which can be directly selected by customers to solve the problems of re-recording and training for such applications, high project start-up cost and long development cycle. It can also support countries around the world to do local wake-up in their mother tongue, and more widely serve as a voice interaction portal for AI hardware, helping domestic electronic enterprise products to go to sea, in addition to Europe and the United States, there is a broader global market.

Whale rhinoceros: HKUST Xunfei, Mobvoi, Plaud.AI and other enterprises have been deeply engaged in the field of AI recording pens for a long time. How do you think there is still much market space in the field of AI recording pens? How to occupy the market share?

Zhu Yuhao: We have done research on the traditional recording pen market, probably less than 20 billion.AI-enabled recording pens have enlarged the original market, such as records of more scenes such as customer visits, meeting calls, classroom learning, doctor-patient communication, etc. This has brought about a new increment, and the total amount may exceed 50 billion or more.

The recording pen is only a microcosm by AI. In fact, it is gradually evolving into AI portable assistant, which can enter all kinds of intelligent terminals, and the market increment space will be larger.

AI portable assistant is an important product line of AI Ling for work and study scenes. In addition to the full range of recording pens, we will also apply them to headphones, watches, glasses and other large items to further expand our market and enable more people to enjoy the efficiency improvement brought by AI.

Whale rhinoceros: It was established in Plaud21. Up to now, no one in China is comparable to it. What do you think?

Zhu Yuhao:First of all, it is related to the market size. As we talked about earlier, the market size of the original recording pen category is not very large, so the big factory did not jump in. The big factory is also looking at a market with at least 10 million shipments. However, it is technically difficult for small companies to do it. AI recording cards are not like standard products for electronic products. Chip companies will provide standard HDK and SDK. The solution companies will do it quickly and, of course, will easily bring homogenization.

AI recording cards have many technologies that need to be re-developed. So far, apart from Plaud, there are not many other similar products on the market that are really mass-produced and easy to use. Ai Ling Technology closely surrounds the demands and promotion suggestions of users of Plaud products in the market, and combines its own underlying technology to achieve a comprehensive upgrade of the ten core functions, bringing better experience to users. I believe we will promote the more positive and positive development of AI recording cards in the industry.

Whale rhinoceros: What are the advantages of AI Ling's products compared with Plaud's AI recording card?

Zhu Yuhao:First of all, in terms of thickness, Plaud's is 2.95mm, and we have achieved 2.89mm, which is the limit of the current industry. More importantly, magnets and wireless charging are integrated. Then we made many optimized designs on WiFi, making the recording data upload several times faster.

At the same time, we also creatively canceled the recording switch button, which simplified the interactive experience. The introduction of intelligent high-performance recording, save storage space, improve the efficiency of computing power; Support recording label function, record important moments and wonderful moments; Integrated translation machine function, support a variety of translation scenes. In addition, Ai Ling has also expanded more categories, allowing AI assistants to enter more scenes.

Whale rhinoceros: What is the current market layout of the company, what are the main target customer groups? What are the characteristics of different countries or regions?

Zhu Yuhao: I hope 70% of the market layout will be overseas and 30% at home.There are two main target customer groups for Ai Ling. One is the global big brands, such as mobile phone/computer brands, which have their own user groups. It also includes the traditional big brands turning to the smart track. The second is the head brands in various regions, including innovative hardware, e-commerce channels, sub-industry applications, etc.

Globally, the user needs of different countries and regions are also different. For example, Europe and the United States are concerned about user privacy, the product life cycle and frequency of use in Japan, and the demand for offline in Africa and other regions.

Whale rhinoceros: The company simultaneously lays out three scenes of work and study, emotional value and comfortable life. What is the specific cooperation mode between Ai Ling and customers?

Zhu Yuhao:We have built a "1 3 N" product system: 1 is the end-side AI technology base; 3 is to focus on three core application scenarios; N is a variety of specific product forms and ecological cooperation. Customer cooperation can deliver both the whole machine and the module kit.

For key scenes, we will directly make the standard products of the whole machine, make the scenes thoroughly and make the products refined. For example, the current cooperation mode with a domestic head mobile phone manufacturer is to deliver the whole machine. For some derivative applications, we will also deliver module kits, open interfaces, and leave more application space to customers and partners.

quality of communication products, cost of consumer products

Whale rhinoceros: Since the product advantage is so prominent, have you ever considered doing ToC directly?

Zhu Yuhao:Our entire founding team has more outstanding experience and ability in the B- side. It takes more energy and money to build and invest in the brand in the C- side. We hope to be more stable in the initial development stage. After the company reaches a certain scale, we will also consider whether to move in the direction of ToC.

In addition, in my opinion, ToB is no less than ToC, its increment is larger. For example, the following three categories: first, after a new AI hardware comes out, a group of innovative hardware companies will jump in and do iteration, differentiation and derivation, which will produce effective increment; Second, big brands choose to enter the market after seeing the market prospect of new categories, and they bring their own large-scale user groups. Third, traditional enterprises have turned to AI intelligence, and a certain proportion of their scale also brings increment.These three types of increments are the target market for Ailyworld Technology and the object of technical differentiation for our products.

Whale rhinoceros: In addition to its technical advantages, what is the core competitiveness of Ailyworld Technology when its products go to sea or cooperate with head brands?

Zhu Yuhao:Ai Ling emphasizes differentiation and price ratio. If a product is not differentiated, we will not consider it. After achieving sufficient differentiation, we will enter the market at the right price. Our price ratio strategy is"The same price is not as good as our performance, and the same performance does not give us the price ".

Ai Ling team has more than 20 years of communication products gene, do consumer products, we hope to have the advantage of dimension reduction.Provide high performance and high quality like communication products, but it is still the cost of consumer products.This is also the competitiveness we are building.

Whale rhinoceros: In the face of wearable equipment, intelligent Internet of Things and other different scenarios of personalized needs, the company's "scenario solution" in addition to providing chip products, but also bring what traditional electronic hardware solution providers difficult to achieve the value?

Zhu Yuhao: The first is the scenario scheme.Traditional electronic hardware solution providers are often passive. Most of them follow the requirements and specifications proposed by customers, and more are to provide technical and human services.

Ai Ling will go deep into the solution for specific scenarios. We will analyze the current industry's pain points, the shortcomings of benchmark competitors, and provide Ailyworld's technical advantages, solutions and product definitions to customers.What we provide is a complete set of solutions with differentiated competitive advantages, which is a weapon for customers to win in the market competition.For the key scenes, we also do very well. Take AI portable assistant as an example. Ai Ling has more than 8 product forms, which basically cover the mainstream needs in the market. Customers can have a wealth of choices and are easier to produce.

The second is technological differentiation.Traditional electronic hardware solution providers generally do secondary development on chip HDK and SDK, which will lead to similar and generally homogeneous solutions supported by this chip company. And Ai Ling is using its own algorithms and chips, based on its own underlying technology to do scenario-based solutions,Ailyworld Inside is a trademark of our technology. What we want to do has its own uniqueness and threshold, and is not easy to be copied and copied.

The mission of Ailyworld Technology is to integrate AI into more scenes and make products more intelligent and simpler.We are constantly looking for more valuable combinations of AI and scenarios, and making this process easier to implement.

Whale rhinoceros: Is there any financial pressure on Ailyworld Technology now? Will there be financing needs (industrial financing or financial financing) afterwards?

Zhu Yuhao:In fact, we choose to start a business at this point in order to build our own platform and pursue a career that carries our values. Compared with young people starting businesses, our advantage is that we have more experience and capital savings. There is no problem in supporting the company's short-term survival and development for a certain period of time. In addition to our own funds, according to our current customer order scheduling and market promotion, it is expected that the company will be profitable in the first year of its establishment.

However, the AI hardware market is flourishing and there are many opportunities. If there are excellent investors or investment institutions in related industries, which will help us increase R & D investment, introduce outstanding talents, and quickly seize market share, we are very happy.

(The author of this article, Xi Xing, is welcome to add WeChat if you are interested in AI hardware.Nebula-yeeexchange of needed goods)