According to research from Accelera, agentic AI systems can increase token consumption per task by up to 100 times compared to traditional AI approaches. The finding highlights potential cost implications for organizations implementing autonomous AI agents that require multiple iterations and decision-making steps to complete tasks.