Sega Cheng, founder of iKala, specializes in artificial intelligence technology, and with the close cooperation between iKala and Google Cloud, they are currently making an all-out effort to push iKala to become Taiwan’s AI marketing and advertising technology expert.
AI and the cloud have been popular technology trends in recent years. There are many examples of enterprises that have improved their operational efficiency because of these new technologies. However, not many enterprises have changed their business models because of these technologies. iKala, however, was one of the few that did.
iKala originally started as a platform that provided streaming services. However, having recently received a new round of financing of US$10 million, iKala has in recent years evolved into a B2B enterprise that provides cloud and AI technology. The company has two major business groups: cloud and advertising technology. These include iKala Cloud that mainly provides Google Cloud services, StraaS that provides various media streaming services, the internet celebrity radar KOL Radar, and the online helper for community sellers Shoplus.
Talking about the company’s history, Sega Cheng, CEO of iKala, indicated that the company first caught the attention of Google because of its use of Google Cloud, as a result becoming one of the earliest global partners of Google.
The three major features of Google Cloud made iKala become a loyal partner without a shred of doubt
When the company just started, iKala was just like most startup companies. In order to establish itself easier and reduce costs, it used the IDCs of telecommunication companies. Afterwards, it moved to AWS; however, the AWS IDC was in Japan, which caused some inconvenience in terms of use.
At the end of 2014, Google Cloud, which had just entered Taiwan, started to take off, and its IDC in Changhua was also set in motion; iKala moved once again. “The main reason was because we worked on multimedia technology and we have high demands for low latency; whether it was for live broadcasting, video downloading or streaming, it was a pretty serious thing if there were delays or lagging,” said Sega Cheng. Having built its own IDC in Taiwan is a great advantage of Google. Unlike AWS, which only has an IDC in Japan, resulting in data transmission being impacted by geological distance, the latency of Google’s IDC in Taiwan is only less than one-tenth or even one-twentieth of that of the Japan IDC.
Another advantage of Google Cloud is its uniqueness in data analysis and storage. Sega Cheng made the analogy that databases are like libraries, the new and popular books are placed on racks that are easily accessible, but the storage cost is expensive; some of the older books are stored deep inside the store room, so the storage cost is low. However, if one day the books in the store room need to be used, the cost to get them out of the store room will be high. At the same cost, the data capture speed of Google Cloud is faster; this is mainly because it uses dedicated networks unlike other cloud providers that have to share the networks with other companies. “The fact that it has low latency and its own dedicated network makes it better than most other companies.”
The third main feature of Google Cloud is even more attractive to iKala: “Its expertise in AI.” Google itself is the greatest technology promoter of various AI applications, especially in areas such as image recognition, voice technology and searching massive amounts of data – Google leads the entire industry in these. Therefore, the use of Google Cloud gives a technical sensation of being on top of the world.
iKala Cloud relies on technology to find business value for customers
Google Cloud not only brought technological convenience for iKala, iKala even found the possibility for business model transformation from Google Cloud. From the beginning when it was only a user of Google Cloud to becoming its distribution service partner in Taiwan, iKala established the iKala Cloud department for dedicated operations.
The underlying services of GCP include transmission and computing, and in the area of application, there are data analysis and the searching of massive amounts of data. Currently, the GCP clients of iKala cover twelve industries across six countries with the main fields including e-commerce, gaming, media, advertisement technology and finance; customers that are more popular in Taiwan include KKBOX, the big game company Rayark, DCard, iCook and CommonWealth Magazine.
Taiwan’s online English-learning platform VoiceTube is a client of iKala Cloud. As an online multimedia platform, VoiceTube faces large numbers of English learners searching for videos online every day, and what they need it a stable and secure platform architecture. In addition, traffic cost is also a cost that must be considered for platform management. GCP provided a huge price advantage in this area.
VoiceTube
As an online English multimedia learning platform, VoiceTube needs a multimedia platform network architecture that has stable transmission and can provide services to the diverse needs of individual users.
GCP can also provide the possibility of a “simple starter AI”. “Google has an API called Video Intelligence, it packages AI technology into AutoML so that some of the things that originally only AI experts could do can now be achieved by average users through simple user interfaces. This means the universalization of AI,” said Sega Cheng. The importance of this to enterprises is that currently when CTOs or CIOs are purchasing IT solutions, they must consider the high shortage of AI talent. According to French statistics, there are only 20 thousand people who can develop AI solutions in the world, and it is neither cost efficient nor feasible for enterprises to hire these people. This is when AI package solutions are needed: “This is the same as liberating AI, and this is a strong selling point.”
Most of the enterprises in Taiwan are small and medium enterprises, which is particularly important. When AI models become AutoML, for one, it allows SMEs to not have to cultivate dedicated AI talent, and for another, they do not need to develop APIs by themselves. This shortens the time for commercial operation. They can try the user interface first, and then finetune it according to their own needs if they find it suitable.
StraaS targets multimedia streaming and traffic management
StraaS, the other cloud business department, mainly targets streaming and multimedia service clients. It has already served Ko Wen-je, the mayor of Taipei, in 2014. “Streaming was provided at many election locations, and once there was an explosion in the number of people online, it became difficult for telecommunication providers to handle the load. At the time, every location for the mayoral candidate Ko almost always had over ten thousand people online simultaneously.” iKala contacted domestic telecom providers, AWS and companies that sold traffic and networks with this data, and it was only when they consulted Google that Google clearly stated that such a scale of traffic was no problem, and it would be high-definition quality. Not only the politics industry, the November 11 promotion of e-commerce also requires stability for large traffic.
StraaS is tightly integrated with the business models of the clients. iKala cloud provides related solutions whether it are subscription systems, traffic amounts or selling advertisements. “StraaS already has these indicators built-in for customers to use as needed.” StraaS specializes in streaming technology, and even the processing of real-time videos and images derived from streaming are also in the service range of StraaS. For example, the AI processing of sales images. Some e-commerce companies have special needs when processing images, having to clear the promotional messages on product images. In the past, the images were cleared manually by engineers, which takes a lot of time; but now with AI, it only takes a few seconds, and the cost is instantly reduced to one-tenth of the original cost.
In addition, in the field of StraaS, iKala also crossed over from only providing underlying technologies in the past to upper layer services and applications. They can now assist clients in whether the customers are satisfied with the services on the consumer-end, how should marketing be done, and whether advertising should be extended with StraaS.
AI trilogy of adult multimedia companies
The latest application case of StraaS is the adult multimedia platform LXAV. LXAV is the only genuine adult video platform that came directly from Japan to Taiwan. It purchased the StraaS system first and then further integrated the operation, maintenance and sales of the website. “LXAV is very classic; they first handled the system aspect and then extended it to the tracking of the sales and website data. StraaS became a one-stop service because of this.”
LXAV
Starting with physical optical discs, this Japanese adult video platform crossed the ocean to Taiwan to establish the only genuine video service platform.
Sega Cheng made the analysis that many customers are actually very unfamiliar with internet marketing. Take LXAV for example, it is an adult video company that started with physical optical discs; it was like all other enterprises that require a trilogy for AI implementation. First is to digitalize the data, second is to establish data analysis capabilities, and third is to implement AI. “This AI trilogy is the same as the process for us to assist our clients through digital transformation.”
Major AI trends of 2019: Traffic pools and the mixing of public and private clouds
iKala helped LXAV move all of its original video files into cloud, then entering the data analysis stage, including consumer subscription status, website tracking, how to advertise and even how to perform marketing with internet celebrities – iKala had complete solutions for them all. Finally, AI models were used to perform data analysis, including where the customers came to the adult website from, which videos they watched, whether they checked out, and whether they searched for the videos with the name of the film, name of the actors or the serial number of the video; these will all affect the design of the website to allow users to have a smoother experience.
When talking about the major future trends of AI and the cloud, Sega Cheng indicated that the main axis is Cloud for Marketing (C4M), with the first trend being the traffic pool. Traditional marketing was based on the intuition of the marketing staff and set budgets with the advertising agencies, but those times are over. Today’s cloud marketing requires more precise analysis. This is mainly because now the acquisition cost of each customer is becoming higher and higher, and enterprises can no longer blindly pursue one-time traffic maximization. Therefore, there must now be in-depth reviews of existing customers – this is the concept of the “traffic pool”. Within this pool of customers, why did some customers come this time but didn’t show the next time? Traffic pools are used to look at the characteristics of users.
The second trend is public cloud plus private cloud. For example, some finance industries require storing some of their data locally due to personal information requirements or company policies; this is when the mixing of public and private clouds is necessary.
Sega Cheng stated that currently, the understanding of AI by domestic enterprises is mostly still at an entry level, and it is already impressive if enterprises can make it to the second part of the AI trilogy – data analysis. There are many aspects where AI can be implemented, including listening, speaking, reading and writing, as well as video images and data analysis, and there are still many more that are worth developing. “In general, the use of AI and the cloud can optimize two aspects for businesses, company operations and customer service, and these are the core spirit of iKala.”