Automating Repetitive Tasks with AI Conversations
Integrating Scattered Data
Reveals Trends, Efficiency, and Insights
Defining the Problem Is More Important Than Coding: CEO Kang of
Says, “We Will Create a
“We will create a dashboard-based communication system”
The culture where marketing department employees routinely worked late into the night to compile ad performance reports has disappeared. Client consultations, which used to rely on heavy paper brochures, are evolving to tablet screens that display the latest case studies with a single tap. Vibe Coding—which allows users to create the programs they need in a flash by conversing with generative AI—is transforming the way Minish works. The driving force behind this change is the frontline staff, who have never studied coding.
Bae Young, Head of Marketing, used VibeCoding to consolidate marketing data that had been scattered across various sources into a single “integrated dashboard.” Previously, advertising data from channels such as Naver and Google was used only to track spending by media outlet, and a significant amount of time was wasted transferring data to prepare weekly meeting materials. With the help of the DX Division, this headache was resolved.
With the dashboard being put to full use starting in late May, we now have a solid basis for decision-making. Not only are advertising costs, impressions, click-through rates, and efficiency automatically aggregated by channel, but we can now track performance in detail down to the ad creative (content) level—something that was previously only possible at the media outlet level.
This has enabled data-driven decision-making, such as quickly filtering out low-performing content and focusing the budget on high-performing content. The dashboard has now been expanded to include hospital-specific revenue, patient referral sources, influencer management, and third-party monitoring, and the performance metrics for YouTube and Instagram content are measured and applied to day-to-day operations.
Director Bae Young-ui said, “Our ultimate goal is to create an environment where we can leave repetitive, routine tasks to AI and focus on essential work such as planning and decision-making.”
Oh Dan-bi, Manager of the TC Team, designed the “Digital Consultation Guide” using VibeCoding. While visiting other dental clinics to observe consultations, Manager Oh was inspired by how clinicians presented clinical cases tailored to patients’ interests. Using “Replit,” an AI development tool, she created a web-based consultation tool that allows clinical cases to be organized according to the situation and easily updated with the latest information at any time, and she implemented the first prototype in just five hours.
If a patient is concerned about the amount of tooth structure to be removed, the system displays cases involving minimal or no removal; if a patient is curious about a specific procedure, it immediately scrolls through relevant clinical cases on a tablet screen to provide an explanation. Development of the functionality is complete, and the team is currently populating the system with content in preparation for its actual implementation. Director Bae Young-ui supplemented the system with Claude code to ensure stability.
This experience also led to the “TC Team Recall Management Program,” which provides step-by-step follow-up for customers who did not make an appointment or return for a visit after a consultation. The system automatically flags customers who are easily overlooked—even when the assigned staff member changes—thereby preventing missed follow-ups.
The two agreed that the key to creating a dashboard lies not in coding skills, but in “how specifically you define the problem you’re trying to solve.”
Director Bae said, “As long as you have a clear plan of what you want to do and the necessary data, AI will handle most of the rest,” adding, “While AI can present various alternatives, important final decisions—such as those regarding the budget—ultimately rest with humans.” Manager Oh said, “Anyone who knows how to use ChatGPT can get started,” adding, “The process of improving tasks one by one, starting with small ones, was a lot of fun in itself.”
This change, driven by the frontline staff themselves, aligns with the direction CEO Kang Jeong-ho envisions for the company. At a recent management meeting, CEO Kang stated, “Rather than simply displaying raw Excel figures, we must create dashboards that anyone can intuitively understand once the frontline staff have fully grasped the data,” adding, “We will complete a system where all operations are managed through AI and dashboards by the end of the year.”
