Crowdworks, 'Workstage' Solution for Major Bank AI Business

A major domestic bank, aiming to lead the era of artificial intelligence (AI) amidst a rapidly changing financial environment, has faced a significant challenge. In the face of the task of securely and accurately linking vast document data with LLM (Large Language Models), AI data specialized company CrowdWorks' 'Workstage' has presented a solution. To meet the paramount values of security and accuracy unique to the financial sector, CrowdWorks' proven data construction capabilities, innovative technology, and stable execution ability were highly praised. Through this large-scale AI project, the bank is expected to transform tens of thousands of unstructured data points scattered within the enterprise into systematic training data assets and continuously secure 'AI-ready data' that AI can immediately understand and learn from. CrowdWorks designs an optimal pipeline for generating data for search and training, thereby building a training data management system that provides integrated support for systematic collection, processing, inspection, storage, and version control of AI model training data. This goes beyond simple data management; it is the work of maximizing the value of hidden data to transform it into a core asset for the enterprise. 'Workstage' is an on-premise based all-in-one data construction platform that consolidates CrowdWorks' long-accumulated know-how in data collection and processing. Its biggest strength is the ability to efficiently handle the entire process from data collection to processing and inspection within the bank's internal security network, without concerns about external leakage and without requiring coding knowledge. The no-code based intuitive work environment helps field staff easily manage and utilize AI training data. Powerful automation features based on OCR (Optical Character Recognition) and LLM are added to maximize work efficiency. In particular, the 'Markdown Editor' feature, which efficiently supports the construction of LLM training data, is revolutionary. By simply designating a box area within a document, text can be automatically extracted, and spelling, spacing, and table corrections can be done with a single click, opening new horizons for LLM training data construction. Kim Woo-seung, CEO of CrowdWorks, emphasized that "securing high-quality data is an essential prerequisite for successful AI adoption," and stated that this collaboration with a major bank has proven the competitiveness of their solution at the highest level in Korea. He conveyed his ambition to continue being a strong partner for companies to securely and efficiently acquire AI-ready data. Workstage is expected to become a core foundation for accelerating AI innovation in various industries, beyond the financial sector.