S2W, led by CEO Seo Sang-deok, has revealed effective examples of how artificial intelligence (AI) can be efficiently adopted by enterprises through 'data intelligence.' This approach underlines the fact the quality of data is not just pivotal for successful AI integration but also influences the evolution of AI technology at large.
On March 2, 2025, S2W representatives stated, "The notion of data depletion hindering AI development holds true domestically as well. S2W’s strides toward improving data quality and achieving notable results serve as meaningful case studies."
One significant example shared was the establishment of the Hyundai Steel's internal knowledge information platform, 'HIP (Hyundai-steel Intelligence Platform).' Built on S2W's industrial generative AI platform known as 'SAIP (S2W AI Platform),' it emphasizes the importance of data as the key element.
Hyundai Steel possesses vast amounts of data amassed over 70 years; nonetheless, much of it is unstructured and difficult for AI to assimilate. The specialized fields of steelmaking and shipbuilding present unique challenges due to their complex terminologies and the lack of accessible external data.
S2W, through its innovative techniques, reported success to process difficult unstructured big data effectively allowing AI to comprehend challenging terminology used within these sectors. Such accomplishments have been hailed internally at Hyundai Steel as significant advancements toward data innovation.
Founded in 2018, S2W has concentrated on analyzing data from the dark web. This experience led to the development of DarkBERT, a specialized language model aimed at dark web domains, recognized as part of Microsoft’s Copilot for Security since the previous year.
Building on these successes, S2W has recently supplied SAIP to Lotte Members for their new AI service, Segment Lab, which focuses on trend analysis. Scheduled for pilot deployment among Lotte Group affiliates starting the second quarter of this year, Segment Lab is regarded as foundational for automated trend analyses combining member consumption and external news data.
This platform can evaluate behaviors of up to 43 million customers and track product sales statuses. It can also generate various trend analysis reports, providing Lotte Group affiliates with valuable insights for product development and marketing strategies.
Ensuring the protection of consumer data during processing is of utmost priority. S2W has developed data protection technologies certified as 'Innovative Financial Services' by the Financial Services Commission. Central to S2W’s data intelligence is, as stated, 'data protection technology.'
Recent focus points on data safety technologies like Security Guardrails and Role-Based Access Control (RBAC) have emerged as key components for reinforcing the safety and reliability of generative AI platforms. The Security Guardrail functions as a defense system to minimize information leak risks by analyzing training datasets and eliminating sensitive details, thereby enhancing data privacy.
This technology demonstrates resilience against 'prompt injection' attempts, preventing harmful responses through pre-defined rules or filtering systems. Meanwhile, RBAC regulates user access to system resources based on job roles, which can be pivotal for AI agents managing tasks instead of humans.
S2W is advancing its technology, incorporating it within Lotte Members applications. This increased emphasis on data processing technology signals the growing importance of data for AI development, aligning with global trends.
Industry leaders like Ilya Sutskever of OpenAI and Elon Musk have pointed out since last year the dwindling volumes of data necessary for training AI models. Their discussions point out the walls scaling laws hit alongside recent shifts emphasizing inference and cost efficiency, yet the clarion call increasingly is, 'it all boils down to data.'
Beyond merely aiding companies with AI integration, S2W recurrently emphasizes data intelligence as the core element to overcoming technological growth slowdowns. Park Geun-tae, CTO mentions, "If data technologies yield significant results across numerous enterprises’ AI platforms, it could stimulate national-level adjustments to data regulations along with comprehensive technological advancements. Data’s effective deployment and protection will represent the benchmarks of AI competitiveness."
The ramifications of S2W’s initiatives extend beyond just corporate efficiency; they advocate for broad systemic changes within the South Korean tech and business landscapes, fundamentally shifting how data and AI interact.