Is Your AI Agent Actually Reasoning or Just Guessing

We're revisiting this episode of Ctrl + Alt + AI, where host Dimitri Sirota sits down with Joe Miller, Chief AI Officer of Vivun, to explore why enterprise AI needs more than increasingly powerful language models. Drawing on his background in physics and years of developing AI products, Joe explains why knowledge representation, structured reasoning, and company culture are essential to creating AI agents that organizations can trust. As enterprises continue to adopt agentic AI, Joe's insights are as relevant today as when this conversation first aired. Revisit the discussion to hear why AI must move beyond search and automation, why hallucinations remain a challenge, and what leaders should consider as AI becomes a more active participant in everyday work. In this episode, you’ll learn: - How reasoning and knowledge representation shape the future of AI agents - Why cultural fit matters for the successful adoption of intelligent assistants - What enterprises should expect as agent technology matures Things to listen for: (00:00) Meet Joe Miller (01:42) Physics background shaping an AI career (04:44) Why presales roles inspired Vivun’s creation (05:29) GPT-3 surprises and the UX breakthrough (08:37) GPT-5’s plateau and persistent hallucinations (12:32) Moving from a small pond to a big ocean (15:08) How Ava supports the full pipeline (21:38) Culture as the differentiator for AI agents