On 24 October, at the invitation of the Korea Industrial Intelligence Association (KOIIA), CLEVI CEO Lee Hwan-ho delivered a keynote speech at the ‘2025 Academy for Developing Key Talent in Mid-sized Enterprises: Future Industry Response Strategy Seminar’.
We participated in the ‘AI Technology Adoption Cases and Practical DX Strategies’ session on the topic of ‘Physical AI Use Cases, Failure Factors During the Introduction of RAG Agents, and Improvement Strategies’.
As interest in AI agents has recently increased, more and more companies are considering adopting AI. Many organisations and businesses are already piloting and operating AI systems.
However, unfortunately, surveys have found that many businesses and organisations either fail to achieve the desired ROI, or that features which worked during the PoC fail to provide proper answers or operate correctly in the main project or at the actual site.
Because of this information, many companies are reluctant to adopt AI hastily.
During the lecture, CLEVI CEO Lee Hwan-ho explained in detail the limitations of RAG systems, a major cause of hasty AI adoption failures, as well as open-source models and small LLMs (SLMs), and explained why previous adoption cases had failed.
Ultimately, the reason for using AI is the expectation that it will perform people’s work for them or deliver better performance than people.
As practitioners at companies that have adopted AI today cannot trust the quality of AI responses, complaints are emerging that their workloads have actually increased because they have to check the AI’s responses one by one.
The reason lies in the fact that most companies fine-tune open-source SLM models and use them as solutions.
Intuitively, one might think that AI will become smarter if it learns something new. However, teaching AI new knowledge instead causes it to lose memories from the past—in other words, arbitrary parameter weights.
An even more serious problem is that no one knows which information or knowledge has been lost.
As adopting AI is not something that requires only a small amount of resources, you will likely have carried out extensive validation from the PoC stage.
Why, then, did these issues not appear at that stage? This was, of course, because it had been retrained using targeted information.
You taught an AI that only knew numbers to learn the alphabet, and then only asked it questions about the alphabet, so naturally it performed well. However, in the field, do we only need models specialised in such specific domains?
Our operational environments face new surroundings and new data every day. However, AI trained by randomly destroying its existing knowledge cannot guarantee either general applicability or accuracy.
Therefore, it is necessary either to possess a From-scratch Foundation large language model or to have the technology to create one.
As these models possess almost all the world's knowledge, they offer not only high-quality responses but also high accuracy.
For specialised domains, you can overcome the shortcomings of a fine-tuned SLM by using a technology called SFT(Supervised Fine-Tuning), or, for small amounts of data, a RAG system or CLEVI's proprietary intelligence database system, Semantic Database technology.
The government also recognises the importance of these foundation models and is providing substantial support.
This is the National Flagship AI Project. The primary objective of this project is likely to secure digital sovereignty.
CLEVI also understood the critical drawbacks of using open-source models, so we developed our models independently, from data collection through to architecture design, and trained them accordingly.
CLEVI's models currently deliver performance that is in no way inferior even when competing with the world's leading AI systems.
In particular, because we possess our own models, CLEVI is not affected by the dependency issues that can arise when using external models.
AI adoption is now becoming a necessity rather than a choice. However, simply adopting the latest technology makes it difficult to achieve the desired results.
True innovation only begins when you accurately understand the limitations of technology and establish the optimal strategy for your organisation's environment and goals.
With differentiated technological capabilities, including its proprietary large language models and Semantic Database, CLEVI is minimising the trial and error and risks that companies experience during the AI adoption process.
We will continue to listen to the voices of those working on the front line and do our utmost to become an AI partner that more companies can trust. We hope you will create the true success of AI adoption together with CLEVI.