Agentic AI: Dissecting the Future of AI Workflows | Memra's Founder Amir Behbehani

This episode explores the transformative impact of agentic AI on software engineering and broader work structures. Against the backdrop of increasing bot-generated pull requests (currently exceeding 13%), the conversation delves into the capabilities of agentic AI, contrasting it with Robotic Process Automation (RPA) by highlighting its reasoning capabilities powered by LLMs and access to long-term memory. More significantly, the discussion examines the shift from automating repetitive tasks to creating "agentic employees" that handle complex workflows, raising concerns about the "black box" nature of AI but emphasizing the improved auditability of agentic AI's decision-making processes. For instance, the ability to trace an agent's steps and reasonings offers greater transparency compared to traditional LLMs. As the discussion pivoted to the future of work, the potential for vertical integration in AI-first companies and the need for a paradigm shift from rigid roles-based economies were highlighted. This means for software engineers a move towards a more managerial mindset, overseeing and managing AI agents rather than solely executing tasks. Finally, the concept of a "synthetic marketplace," where work is described and agents are allocated accordingly, was introduced, suggesting a future where human-AI collaboration will redefine the nature of work and labor contracts.