AI data company IGAWorks said on Aug. 28 it has built its in-house AI foundation model, WorksFM, based on real consumer behaviour data covering 34 million people.

Through WorksFM, IGAWorks will move to develop the Consumer LBM (Consumer LBM, Large Behavior Model) field, aimed at understanding consumer behaviour and inferring future actions.

The company said that while LLMs (Large Language Models) learn vast language patterns to predict the next word and sentence, LBMs learn continuous real-world human behaviour data to infer current states and the likelihood of subsequent actions.

WorksFM focuses on learning consumer behaviour flows and shifting patterns that occur across different areas, including app use, payments, offline visits, ad responses and TV viewing.

IGAWorks unveiled Synthetic Consumer Intelligence, or SCI, along with the technical architecture and application strategy for WorksFM, at the Snowflake World Tour Seoul 2026 event held on Aug. 27. Shim Joon-hyuk (심준혁), CBO of IGAWorks, presented how the company has converted the vast data assets accumulated over the past 20 years into a behaviour data structure that AI can learn from. He also outlined a strategy to connect it to companies' actual decision-making through SCI and WorksFM.

Shim said, "We are currently working on projects with 2 major domestic retail companies to combine SCI with the data each holds." He added, "We will expand a 'data alliance' across various industries by linking IGAWorks' behaviour prediction technology to companies' internal data and evolving it into more useful and advanced assets."