The upgrade from flat fees to variable pricing is designed to support the use of AI agents for LLM inference, compute and data queries.