The hosts present a three-level AI fluency framework: Level 1 uses AI as a simple assistant for tasks like writing emails; Level 2 involves creating automated workflows for repetitive processes, such as generating action items from meeting transcripts; and Level 3 employs autonomous agents that independently execute complex goals, like building a website. They emphasize that many people dismiss AI after a single bad experience, but the technology improves daily. The key to overcoming skepticism is finding a personal 'wedge'—a practical, time-consuming task where AI can demonstrate immediate value. Leaders should show, not just tell, by visually demonstrating AI performing a specific task and then letting the person try it themselves. The hosts also warn against measuring AI adoption by token spend or using it to optimize flawed processes, advocating instead for viewing AI as a people enhancer. They recommend advancing through the three levels, starting with the keyword 'skill' for workflows and progressing to autonomous agents with tools like Codex.