[SeoulTech x Dong-A.com Joint Planning] Seoul National University of Science and Technology (hereafter referred to as SeoulTech) operates various support programs to assist the growth of startups, including preliminary and early startup packages, maker spaces, and global collaborations. Furthermore, in partnership with Dong-A.com, it provides global news to aid the overseas promotion and expansion of startups, introducing promising deep tech startups both domestically and internationally.


Introducing AI services to enhance corporate competitiveness is now more of a necessity than a choice. However, the process of adopting AI services within companies is not straightforward. Companies often face difficulties due to differences in employees' AI utilization capabilities, and implementation can be thwarted by security issues. If employees have varying abilities to write AI prompts, resulting in inconsistent quality of outputs, or if there is a risk of confidential information leakage while using external AI services, it poses a significant problem.

‘WYHIL (CEO Hunter Shin)’ is an AI platform company that addresses these issues. Starting with Internet of Things (IoT) services in 2015, it now provides a generative AI-based business platform, offering an integrated AI service solution that simultaneously enhances AI utilization and security for companies. The press interviewed CEO Hunter Shin to learn about the various aspects of their AI platform technology and their future vision.

- It is heard that you worked at various large corporations domestically and internationally before founding the company in 2015. What motivated you to start a business after leaving a stable career?

: Originally, I was an engineer and a technology consultant. While working at Oracle Korea and LG Electronics, I found that the biggest issue was the lack of cooperation between the existing electronics and software industries. Decision-making was particularly slow. Having experienced both the field and management sectors, I thought about solving this problem.

Initially, I started with IoT technology, wanting to create new solutions by integrating IoT, electronics, and software technologies. Later, I also dealt with AI, and my field experience and know-how were very helpful in this process.

- Why did you shift your business focus from IoT to AI in the early stages of your startup?

: We actually started with IoT initially, but now we have shifted the company's core capabilities entirely towards AI, particularly generative AI. However, we have not completely abandoned IoT. This is to prepare for the upcoming era of physical AI (robots, etc., AI in the physical world). We are continuously testing the integration of IoT infrastructure and AI in the edge domain.

- WYHIL emphasizes providing various technologies and services for incorporating AI into business. Could you explain in detail?

: Our company's main product is a platform that integrates various existing AI-related technologies into one. The platform is called ‘crux,’ which is the scientific name for the Southern Cross. The Southern Cross provides the correct direction when navigating the vast sea. We named it so to symbolize the same in the AI field. The core solution is ‘crux:SINGLEPOINT.’


The feature of this solution is solving security issues that arise when existing companies use external services like ChatGPT or Copilot. The concept is to manage various related processes at a single point (SinglePoint).

While developing AI agents to automate various corporate tasks, we discovered a problem. The skill or level of using AI varied greatly among employees. The results differed significantly depending on which language model was used, how the prompts were created, and what reference data was provided to the AI.

For instance, when reviewing contracts, using a low-performance model could completely misinterpret the intent of the contract. Relying on each employee's AI skills to proceed with tasks posed too much risk.

Therefore, within crux:SINGLEPOINT, we included an appropriate level of models, optimal prompts, and essential data so that all employees could use this system to produce AI outputs of uniform quality. The key is to standardize or optimize employees' AI utilization capabilities.


- It is said that the crux platform can be expanded in various ways. What other notable features are there?

: The most core AI technology is RAG (Retrieval-Augmented Generation). It allows for AI-based search, document creation, etc., based on the data or information a company possesses. For this, vector search and similarity search technologies are needed, and we have created a product called ‘crux:SEARCH’ that extracts only this function.

This provides extended functionality beyond existing index or character-level searches and keyword matching-based searches. It analyzes the query content to display similar results, and multimedia searches such as video, image, and voice are also possible. This technology greatly aids the digital transformation (DX) of corporate business overall.

- Cloud environments are advantageous for convenience, while on-premises environments are advantageous for security. What environment does crux support?

: Installation is possible for both, but we are currently focusing on the on-premises environment. The goal is to install it independently within a company's security network and internal network.

There is a lot of concern about data leakage or information leakage when using AI, but crux can implement everything, including language models, platforms, and data processing, within the company without using the cloud. It can be seen as a product with enhanced security. Recently, many companies seem to prefer on-premises to eliminate potential issues.

Of course, we have already developed means to allow our service to be linked with external global services. Therefore, it is possible to connect and use external language models as needed. The characteristic of the crux platform is considering security while adopting advanced AI technology.

- Are there any cases where it has been applied to actual companies or institutions? If so, what was the response?

: Our main targets are public, financial, and educational institutions. They share a common reluctance to adopt cloud-based technology due to security issues.

Among educational institutions, we are supplying AI for expert search and specialized information search in collaboration with the Chosun University RISE Business Group and industry-academia research linkage. We also provide a business platform for business advancement of the business group.

Besides that, we are currently conducting many PoCs (Proof of Concepts) in various sectors such as public, financial, and logistics. We received a good response from the Credit Guarantee Fund and ranked first in performance evaluation compared to competitors in the MRO AI service. We also conducted a PoC at YTN broadcasting company, which received positive feedback. These experiences will greatly aid future business expansion.

- Competition in the AI market is very fierce. What are the representative differentiating factors?

: The most differentiating factor is the RAG-related technology. We have secured many modules through patents and self-development. We have secured many unique technologies, especially in the process of acquiring data for RAG and processing this data.

In addition, we have significant capabilities in vector database management technology. This greatly affects the overall AI service system, including the quality of responses. Most of the first-place results in PoCs were due to vector data management technology, and we received high evaluations, especially in resource optimization.

- Small startups have to overcome even tougher competition. How are you overcoming it?

: Although we have been technically recognized in PoCs, there were naturally concerns due to our company's size, revenue, and number of employees. Therefore, we are strategically pursuing OEM/ODM partnerships to allow partners larger than us to supply our technology. We are currently in partnerships with well-known companies like Company H and Company E, supplying products. The technology is ours, and the sales are handled by the partners.


- What are your impressions of using SeoulTech's global collaboration program?

: We were selected for the 'Miracle' program in collaboration with Oracle in this project. This Miracle was our last chance. After more than 10 years of establishment, we cannot receive support programs.

Thanks to this, we received various technical collaborations. In particular, we were able to learn and test the vector management technology know-how of Oracle DB in relation to vector data. SeoulTech provided equipment for us to test in an on-premises closed network, and Oracle provided an environment to run our system based on Oracle Cloud.

In addition to technical support, we are also greatly benefiting from the partnership with Oracle, a global company, in terms of overseas market entry.

- It seems you are also attempting to expand overseas. How far has it progressed?

: We are working hard to expand overseas directly. In particular, we are continuously requesting cooperation for expansion into Japan and the United States through Oracle. A visit to Oracle's Japan branch is scheduled for December this year.

We plan to approach overseas business in the form of SaaS (Software as a Service). Collaboration with Oracle, which has recently transformed into a cloud company, will be of great help. We are also exploring Southeast Asian markets such as Malaysia and Indonesia, targeting countries that do not directly develop AI.

- What are your future plans and aspirations?

: We expect the public, financial, and education markets to expand further next year. There are many requests, especially in the education sector, and we believe it is important to enhance understanding and utilization of AI in this field.

We are currently undergoing GS certification and procurement registration procedures to supply to public procurement. As the national government is actively promoting AI adoption in public institutions, this market seems more attractive next year.

Although we are small in size, most of our employees are engineers with excellent know-how. I am still coding directly. Today's AI is different from tomorrow's AI. Therefore, if you need an AI partner to grow with customers, please contact us.

IT Dong-A Reporter Kim Young-woo (pengo@itdonga.com)

Popular News