Businesses are increasingly looking to automate human tasks for greater efficiency, and IBM’s latest research focuses on using generative artificial intelligence (AI), such as large language models (LLMs), to improve automation. The proposed software framework, SNAP, trains an LLM to predict the next action in a business process, offering suggestions for the next steps. This approach leverages semantic stories and natural language narratives to capture more details of a business process, going beyond the capabilities of older AI programs. The researchers found that SNAP outperformed older AI programs in next-action prediction, particularly when the dataset contained rich semantic information.