[Forbes Korea, May 2025] Park Ji-hyeok, CEO of Waddle, and Kwon Eui-jin, Director of Dalpha — AI Is Merely a 'Means,' Not a Goal

In the May 2025 issue of Forbes Korea, Waddle CEO Park Ji-hyeok and Dalpha Director Kwon Eui-jin held a conversation on how to leverage AI agents. They discussed the AI agent boom, future outlook, and points of concern. You can read the full text below.

Park Ji-hyeok, CEO of Waddle, and Kwon Eui-jin, Director of Dalpha — AI Is Merely a 'Means,' Not a Goal

"AI agents are a means, and they must not become the goal." Two young founders running AI agent startups spoke with one voice. They explained that even though AI agents boost the efficiency of corporate operations, companies should not adopt them without a clear sense of purpose. Park Ji-hyeok, CEO of Waddle, and Kwon Eui-jin, Director and co-founder of Dalpha, agreed that the focus should be on essential value rather than the profitability of the technology.

It was startups, not Big Tech, that led the early AI agent market. When the AI boom erupted with the arrival of OpenAI's ChatGPT in late 2022, the attention of most Big Tech players turned to generative AI–related technologies. The main subjects of research were large language models (LLMs)—which learn from massive amounts of data to understand and generate human language—and multimodal AI, which simultaneously processes various forms of data such as text, images, and voice. In particular, the number of parameters (the count of individual parameters included in a model) was the key metric in the LLM race.

While the parameter race was in full swing, quite a few startups threw in the towel. Taking on Big Tech in LLM development—which requires enormous capital and manpower—was a daunting challenge for startups. Even so, the choices made by a few AI startups hit the mark and proved to be a 'masterstroke.' They moved into the AI agent market, which was not even booming at the time. According to global market research firm Grand View Research, the AI agent market was worth around $3.8 billion in 2023.

One factor behind so many startups jumping into AI agent development was the prolonged investment winter. With funding drying up, AI startups searching for technologies with relatively low entry barriers discovered the 'oasis' of AI agents. At the time, several Big Tech players such as Meta, Google, and IBM released their AI models and application programming interfaces (APIs) as open source, allowing startups to accelerate their AI agent research and development (R&D) on that foundation.

Another masterstroke by AI startups was choosing enterprise-tailored models as their business item among the countless types of AI agents. B2B (business-to-business) ventures have the advantage of being more stable than B2C (business-to-consumer) ventures, because securing long-term contracts with key clients can guarantee profitability for a certain period. As cases of enterprise AI agents actually boosting work efficiency and productivity gradually increased, the AI agent market in 2024 (about $5.4 billion) recorded a growth rate of roughly 40% over the previous year.

AI agents were nothing less than an innovative weapon that allowed companies to outpace competitors. As client demand grew by the day, AI startups developed AI agents specialized for various fields such as healthcare, finance, education, and customer service, along with hyper-personalized services. The bold challenges and swift innovations of startups brought the era of AI agents closer. Grand View Research forecast that the AI agent market will grow at an average annual rate of 45.8% over the next five years, reaching $50.3 billion by 2030.

Having witnessed this phenomenon, Big Tech has become convinced of the rosy future of the AI agent market, and the technology development race appears to be intensifying. Not only Big Tech but also telecom carriers and information technology (IT) firms are hastily joining the market under the banner of being AI companies.

AI Agents Are a 'Genre' of Their Own

As the battle for dominance over AI agents deepens, the concerns of domestic AI startups appear to be growing as well. The deeper their anguish, the faster they are striding toward advancing their technology. We sought out two domestic startups fighting hard on the global AI agent battlefield: 'Waddle,' which develops and supplies the conversational AI shopping agent Zigzag (Gentoo), and 'Dalpha,' which provides enterprise-tailored AI solutions as software-as-a-service (SaaS). Within the industry, both startups are recognized for being well-equipped with AI agent technical capability and marketability alike.

Waddle, founded by CEO Park Ji-hyeok in 2019, took note of the phenomenon where online shopping mall visitors leave for search portals, social media, or YouTube to resolve their questions about products. Judging that having a clerk in an online mall too could prevent potential shoppers from leaving, the company developed Gentoo, an online clerk. Gentoo is praised for excelling at recommending products to suit customer needs, situations, and context. Last October, OpenAI signed an enterprise contract with Waddle—its first with a domestic startup.

Dalpha, which helps companies automate their work, is an early-stage startup founded in 2023, but it boasts a rapid growth pace. As of this past April, the number of clients that have adopted Dalpha's AI solutions—which include enterprise-tailored AI agents—reaches about 160. These clients span diverse industries and sizes, including Hyundai DF, KT Commerce, Daesang, SK Stoa, Daehong Communications, Amorepacific, and Makeus (operator of Dingo). In early April, at the JoongAng Ilbo Building in Sangam-dong, Seoul, we met Waddle CEO Park Ji-hyeok and Dalpha Director and co-founder Kwon Eui-jin to discuss the AI agent boom, future outlook, and points of concern.

How do you view the AI agent boom?

Park Ji-hyeok (hereafter Park): Many domestic IT startups have transformed into AI startups amid the AI craze. Among them are companies that build AI models, some that design the semiconductors AI requires, and others that run platforms using AI. Countless types of startups have branched out from the new technology of AI. Now AI has moved beyond being a specific technology and has become a genre of its own. I think AI agents, too, will develop into yet another genre. That's because the concept of AI agents is evolving like a spectrum. Vague anticipation has passed through a phase of concretization and entered an era of technological advancement. Just two or three years ago, AI agents remained merely a matter of curiosity. But as more companies use AI agents to boost productivity, their requirements are becoming more concrete as well.

Kwon Eui-jin (hereafter Kwon): When I look at the recent AI agent boom, I'm reminded of the 'dot-com bubble' of the 1990s. The bubble economy brought on by the popularization of the internet at the time faced a crisis in the form of a stock market crash, and countless companies went bankrupt one after another. I worry that this dark chapter of history might repeat itself. Whether AI companies, clients, or ordinary users, there's a common tendency to blindly worship the new technology of AI. But what happens if the hopeful expectations surrounding AI turn out to diverge somewhat from the reality to come? We need to realistically temper our excessive expectations of AI while at the same time considering how to put AI to constructive use.

Is there an example that shows blind worship of AI?

Kwon: Quite a few companies treat the adoption of AI technology itself as the goal. They're filled with the belief that 'our company too will grow faster if we just bring in AI agents or AI systems and use them.' But AI is not an 'all-purpose item.' AI is merely a tool, and the first priority is to grasp the company's goals and immediate challenges concretely and meticulously. Only when the problem is clear can the right tool for solving it be properly utilized. A company that tries to use AI technology blindly, without a clear sense of the problem, will struggle to create real value. Adopting AI technology is, for a company, nothing less than an investment for the future. Since it incurs enormous costs, we must guard against blind worship.

But the position of AI companies would seem to be different.

Kwon: Of course. The position of technology companies that develop various AI solutions, including agents, is bound to differ from that of clients who use those solutions. Yet the two can be called a community bound by a shared fate. That's because if a client's sense of the problem isn't clear, the technology company can't guarantee profitability either. If you adopt AI technology right away without any preparation, you may gain short-term utility, but it's hard to achieve steep, rapid growth over the long run. Even linear growth becomes impossible. That's why technology companies like Dalpha must help clients focus on essential value rather than chasing illusions. This is both Dalpha's reason for survival and my personal wish.

Unconditional AI Adoption Should Be Avoided

Voices warning of the dangers of AI agents are also growing louder.

Park: AI has the drawback of being unpredictable, because the same input can yield different results. This is both a distinction from conventional software and a risk factor. In the e-commerce domain in particular, AI agents can induce consumers to make mistaken purchases through incorrect information, ambiguous explanations, or time lags in product updates. If such risks recur, they can be fatal to corporate operations. Companies that use AI agents must build systems that can identify various error types and respond swiftly. Moreover, the ripple effect of AI is not confined to companies. To prevent its social influence from expanding in the wrong direction, national-level guidelines should be established.

How do you plan to respond to the criticism that the locus of responsibility is unclear?

Park: As the number of people using AI agents increases, the issue of where responsibility lies will become even more sensitive. For example, if someone misuses an AI agent, the AI agent developer, the online mall operator, the mall's tenant sellers, ordinary users, and cyber attackers are all entangled in the incident, so resolving the problem is bound to be complex. Also, the more new users flow in, the greater the chance that unforeseen incidents involving AI agents will newly arise. We must also prepare for cases where an AI agent takes actions that deviate from the developer's intent and purpose. Considering every possible scenario, we need to carefully distinguish the locus and proportion of responsibility for each individual issue.

I'm curious about your ultimate business vision for using AI agents.

Park: Ultimately, Waddle hopes that more users will use Gentoo to efficiently improve their purchase journey. Our goal by the end of this year is to surpass one million users who have talked with Gentoo over the course of a week. In an offline store, it can sometimes feel burdensome or bothersome when a clerk speaks to you first. But the online clerk Gentoo can naturally carry on a conversation without going against the wishes of a user who is in the middle of shopping. The more cases there are where Gentoo enhances users' purchase convenience, the more online shopping mall stores will adopt Gentoo. On the technical side, going beyond product search and recommendation functions, we plan to provide a notification service so that users can buy the items they purchase regularly at just the right time. Please look forward to Gentoo, which is gradually evolving from a simple clerk into a premium concierge (guide).

Kwon: Among the expressions Dalpha's members like to use, there's the term 'momentum.' Momentum refers to the decisive flow of how things are unfolding. We coined the term 'momentum' to mean we should never forget the feeling we get when, on rare occasion, things are going well. I hope Dalpha will keep moving forward while holding on to that momentum. In the short term, our goal is to attract more clients and expand our B2B business. From a long-term perspective, we want to see more clients who keep using Dalpha's subscription service over the long haul. The indicator that captures this is Monthly Recurring Revenue (MRR), and we plan to work so that our MRR growth rate can climb steeply. We will also shed our fear of the high-risk, high-return strategy and boldly take on the global market. Please keep an eye on us.

Source: Park Ji-hyeok, CEO of Waddle, and Kwon Eui-jin, Director of Dalpha — AI Is Merely a 'Means,' Not a Goal

If reading this article has sparked your interest in Dalpha?

Dalphaprovides, through 1:1 consultation,
AI agents optimized for each company and field— planned and built for you.
If you'd like to learn more about a free consultation,please reach out to us anytime!

Inquire about a custom AI agent for your business