"Restaurants Without Agents Will Disappear"... CEO Yoo Ho-hyun Predicts Survival Strategies in the AI Agent Era

As the era where AI agents become the center of daily life and work approaches, the structure of organizations that embrace the technology and the role of data are emerging as more critical variables than the performance of the technology itself. Yoo Ho-hyun, CEO of Toble AI, analyzed that in the process of preparing for the AI agent era, the more important issue is how to redesign human mindsets and organizational goals rather than the technology itself.

The reason why even if given a truck, it won't be used: The gap between technology and the field

CEO Yoo Ho-hyun explained the reason why the introduction of AI agents fails in the field using a logistics innovation case. Suppose there was an optimized system in the past where 13 people carried goods in a relay method every 100m for a 1.2km section from Gwanghwamun to City Hall; even if an innovative means like a truck is introduced, the field does not change immediately. If trucks are provided to all employees, it may actually take more time during the U-turn and loading/unloading processes than the existing relay method, which only increases costs and leaves utilization low.

Yoo diagnosed that the current situation of AI utilization in Korea is similar to this. Even if AI agents are installed, the situation repeats where people stick to existing work methods (writing documents, coding, etc.) and spend more time correcting the results created by AI. He pointed out that the utilization of AI agents is not simply a matter of technology, but a matter of making people understand and reorganizing the organization's skill set. Just as the introduction of a truck requires more than a simple equipment replacement, demanding a redesign of logistics routes and new personnel composition (drivers and specialized loading/unloading personnel), AI must also be accompanied by the optimization of the entire organization to achieve true efficiency, he analyzed.

The era when App Stores disappear: The arrival of the 'Agent Economy'

A transition from an era where users directly search for and execute apps to an era where they issue commands to AI agents was also predicted. Yoo predicted that just as the past method of ordering by phone moved to the smartphone app-based platform economy, AI agents will evolve into a stage where they receive commands from users and operate apps on their behalf.

In this process, the survival methods of companies and individuals will also change. Yoo said that in the agent era, because agents search for information and perform orders, restaurants or services that agents cannot find will be in a state equivalent to not existing in the market. This means a structure where, beyond the phenomenon of being dependent on platforms, economic activity itself becomes impossible if one is not selected by the agent.

Furthermore, an organizational model was presented where the entire company moves around a single agent, going beyond the stage of individuals utilizing agents. Citing the case of the organization operated by Yoo, he explained a structure where an AI agent is deployed into a group chat to perform tasks such as writing business plans or creating logos, and members make decisions by reflecting opinions in real time. This can act as a new work paradigm that skips existing reporting systems and maximizes work speed.

On-device AI and the importance of data accumulation

Regarding the technical aspect, he mentioned the rise of 'On-device AI,' which runs AI on the device itself without going through a server. The outlook is that if LLMs (Large Language Models) begin to run directly on phones or computers around next year, an environment will be created where AI can be used infinitely without separate server costs (token fees). Accordingly, rather than depending on models of specific countries or companies, competition for services using open-source models or on-device environments is expected to intensify.

However, Yoo emphasized that the difference in agent performance will ultimately be determined by 'data' and 'ontology (worldview).' He argued that an agent only possesses practical value if it understands the business context of a specific company and continuously learns to accumulate experience. A stark difference in quality will inevitably appear between a newborn AI and an AI agent that has accumulated data through numerous corporate consulting experiences, and he analyzed that ultimately, how one injects which data to grow the agent will be the core of future AI competitiveness.

Source: original video (YouTube)