AsiaTechDaily – Asia's Leading Tech and Startup Media Platform
Artificial intelligence (AI) has experienced rapid development over the past year. Consumers have enthusiastically embraced generative AI with open arms, ushering in an era in which everyone can learn AI on their own. However, the enterprise sector presents a different scenario.
In 2023, Silicon Valley venture capital firm Menlo Ventures published The State of Generative AI in the Enterprise The report revealed that enterprises spent approximately US$2.5 billion on generative AI in 2023, fueling the rise of tools like GitHub Copilot and Hugging Face. Despite this, the market remains in its early stages. Currently, enterprise investments in generative AI account for less than 1% of total cloud spending. This indicates that while consumers are rapidly adopting generative AI, enterprise adoption lags, reminiscent of the early days of cloud technology.
Turning to Taiwan, Microsoft’s Work Trend Index report highlights that although 79% of business leaders believe that integrating AI is crucial for maintaining competitiveness, 59% are concerned about quantifying AI-driven productivity gains. Additionally, 60% of leaders admit that their companies lack a clear vision and plan for AI implementation.
This gap is precisely the opportunity that AI transformation solution provider iKala aims to address.
“How is generative AI applied in the enterprise? I believe that knowledge management is a significant area. ” Sega Cheng, cofounder and CEO of iKala, talked about the recent development of AI and pointed out that in the past, most companies hesitated to move forward in digital transformation due to cost considerations and the challenge of measuring effectiveness. “Companies spend time and manpower organizing data and building data pipelines, yet fail to see practical applications, leading them to question the value of digital transformation.”
However, the advent of generative AI offers a breakthrough. Cheng explained, “After the emergence of generative AI, the data pipeline obviously has an outlet, and that is the Knowledge Management System.”
Knowledge Management: The Critical Application of Generative AI in Enterprises
At the end of 2022, generative AI emerged. AI has become a must-learn topic for everyone, and enterprises are gradually seeing the dawn of AI applications. “Over the past year, the most noticeable increase in demand from enterprises has been for knowledge management. We are frequently asked how to implement knowledge management,” Sega Cheng said. He noted that with the advent of generative AI, the industry’s emphasis on data is rapidly growing. “Integrating a company’s exclusive know-how into a language model combined with AI search capabilities will significantly enhance the experience.”
The so-called knowledge management system can effectively document and accumulate a company’s proprietary knowledge, transforming it into a valuable asset for adapting to market changes. For organizations, knowledge management systems are overall resources that can be shared and can help enterprises in decision making, problem solving, future innovation, etc. Leveraging a company’s proprietary knowledge is crucial to boosting competitiveness.
However, many enterprises have faced several practical challenges when establishing knowledge management systems. The most significant issue is data being scattered across various locations, with no proper “data pipeline” established.
According to a 2023 survey by research firm Gartner, knowledge workers use an average of 11 applications, with 40% using far more than the average. Furthermore, 47% of workers frequently struggle to find the data or information they need to complete their tasks effectively. “If the data pipeline is not well connected, subsequent AI applications are impossible,” Cheng said.
Moreover, even when companies successfully build internal databases, poor user experience and difficult-to-use systems are significant barriers to adoption.
Cheng noted, “Anyone who has used traditional library systems knows how challenging it can be to find information. Even if the data is digitized, people won’t use it if the system is not user-friendly.”
Historically, enterprise knowledge management systems lacked interactive search engines, resulting in a clunky and inefficient user experience. Additionally, these systems require continuous support and training to help employees become proficient. Without this support, employees are often left unsure of how to use these tools effectively.
AI to the Rescue! Every Company Should Build Its Own “Enterprise Brain”
AI is undoubtedly a large-scale arms race. The latest report from investment institution Bernstein predicts that in the first half of this year, capital expenditures by major technology companies such as Amazon, Microsoft, Google, Meta, and Apple have reached an unprecedented US$200 billion. This spending is expected to continue increasing, with the majority focused on data centers, chips, servers, and equipment related to training generative AI models.
Sega Cheng emphasized that for enterprises to stay ahead in this AI revolution, they should avoid the costly endeavor of training their own AI models. Instead, companies should focus on infusing their proprietary data into existing foundational large language models, organizing internal information effectively, and leveraging generative AI in conjunction with vertical search to “revitalize the field of knowledge management.”
He illustrated this by comparing ChatGPT to a “generalist” and the vertical AI developed by enterprises to a “specialist.” This specialist acts as the company’s “enterprise brain,” which will help businesses to strategically deploy and gain a competitive edge in the AI era.
With generative AI, the enterprise’s knowledge management system is like a powerful application scenario. The system will be able to sift through vast amounts of unstructured data and combine this information to generate useful output. Employees can effortlessly access the information they need through simple conversational queries.
Cheng remarked, “The advent of generative AI is akin to a previously untapped data pipeline finally finding its outlet. Vertical search within the enterprise represents the most effective application scenario. As natural-language interfaces advance and AI becomes adept at recognizing a company’s proprietary databases, it will herald a groundbreaking paradigm shift.”
KOL Radar: An AI-Driven Knowledge Management Paradigm
Knowledge management is the next market gap that iKala aims to address. But in fact, as early as 2018, when iKala launched “KOL Radar,” it used the “search + AI” architecture, which is the application of knowledge management.
Cheng explained that KOL Radar gathers global influencer data and integrates it into more than 140 language models. Some models analyze audience emotions and preferences, while others tag and profile customers. “These 140 models function like different experts collaborating to make decisions; it’s akin to having a head chef and sous-chefs in a kitchen, together creating a well-rounded and delightful meal.”
This approach exemplifies the “Mixture of Experts” (MoE) in large language models. MoE comprises multiple smaller models (experts). Each expert specializes in a specific task. The system evaluates which expert is best suited for solving the problem, improving efficiency and accuracy, and saving resources and computing time.
Take iKala as an example. Suppose the brand owner seeks out influencers to promote coffee. Searching by keyword may only find influencers who have endorsed coffee. “But AI will not only search, but also associate, to find internet celebrities who are more suitable to endorse ‘lifestyle,’ with a wider scope and a more precise audience.”
The “influencer” domain is a vertical search field that involves not just finding information but identifying the right individuals. These influencers have diverse audiences, characteristics, and preferences that evolve over time. “Coffee connects with lifestyle, and AI generates a knowledge graph that surpasses keyword searches,” Cheng stated. By harnessing AI’s analytical and interpretative capabilities, search results become more precise, aligning seamlessly with enterprise knowledge management—a future direction for companies to embrace.
Over the years, KOL Radar has amassed over 3 million international influencer profiles and billions of real-time social media data points. It is the largest influencer database in Asia, serving more than 50,000 brands and commanding over 70% of the market share in Taiwan. Cheng emphasized that KOL Radar is a prime example of an effective enterprise knowledge management system.
Maintaining Startup Agility and Flexibility
After graduating from National Taiwan University and earning a master’s degree from Stanford University, Sega Cheng joined Google as an engineer. He was also the first speaker in Taiwan to attend the Google I/O Developer Conference. In 2011, he resolutely gave up his stable corporate job and founded iKala, which has now entered its 13th year. Judging by age, iKala is a mature traditional enterprise, but it still retains the flexibility and agility of a new startup to keep pace with the times. From its origins as an online karaoke platform to cloud services, and later developing marketing technologies like KOL Radar and CDP, iKala is now riding the AI wave. At every turn in its entrepreneurial journey, iKala has navigated with precision.
Recently, iKala received more than US$20 million in Series B+ financing led by Chunghwa Telecom. The two parties will launch strategic cooperation to help more companies achieve AI transformation. Additionally, KOL Radar has entered the Japanese market and is growing rapidly. iKala has achieved remarkable results in every business venture.
When asked about the decisions behind each pivotal moment, Cheng humbly smiled and said, “The primary consideration behind these decisions was simply to ‘survive.’” He acknowledged that every turning point on the road to entrepreneurship is crucial and that any misstep could lead to failure. The leadership team needs to have strong psychological quality and willpower.
Looking ahead to the next decade, Cheng believes that AI will revolutionize business operational efficiency. iKala aims to become a leader in enabling enterprises to harness AI, helping them seize opportunities, boost productivity, and accelerate growth in the AI era.
This article is part of a partnership with Cherubic Ventures. Founded in 2014, they are an early-stage venture capital firm that’s active in both the US and Asia, with a total AUM of 400 million USD. Focusing on seed stage investments, Cherubic aims to be the first institutional investor of the next iconic company and back founders who dare to dream big and change the world.