According to Futurum Research, 89% of CIOs (chief information officers) consider agentic AI a strategic priority, and “agent-based AI will drive up to $6 trillion in economic value by 2028, accelerating AI’s role in automating enterprise workflows.” According to Gartner, agentic AI will by autonomously resolving 80% of common customer service issues without human intervention by 2029.
While it remains a foundational solution for task and process automation, Robotic Process Automation (RPA) is strictly rule-driven. RPA vendors have responded to the emergence of large language models (LLMs) to offer Agentic Process Automation (APA) involving the use of AI Agents to automate processes, employing LLMs to improve decision making and execution capabilities. However, APA is still lagging Agentic AI Automation in specific areas. A good case in point is dynamic, non-linear workflows requiring real-time decisioning; APA is only good enough for semi-dynamic workflows that require adjustment based on inputs. Agentic AI Automation can handle low volume, highly-unstructured, or complex tasks (with suitable LLMs); APA cannot meet such requirements.
Agentic AI goes beyond RPA, BPA, APA, machine learning (ML), large language models (LLMs), generative AI, and even AI agents themselves.
Agentic AI automation requires an orchestration layer to improve visibility, reduce complexity, and effectively manage agentic workflows.
With many vendors offering AI agents, business and IT leaders need to be aware that hurling individual agents at isolated workflows is a costly approach that is unlikely to deliver much value.
Without orchestration of AI Agents, enterprises will be adding more complexity to a spaghetti architecture for automation tools.