Ashwin Rao on Half the AI, Half the Answer: Why LLMs Can't Do It Alone | Ep 1284

In this episode of The Digital Executive Podcast, host Brian Thomas sits down with Dr. Ashwin Rao, Executive Vice President of Next Gen AI at o9 Solutions, to explore why large language models alone are not enough to power the autonomous enterprise — and what it will actually take to get there. Ashwin opens by tracing the common thread across his career on Wall Street, at Stanford, and in retail supply chains: every domain ultimately comes down to translating messy business problems into clean mathematical models, then building software to solve them. He then makes a compelling case for neurosymbolic AI, arguing that LLMs represent only half the picture — the neural, statistical, approximate half that excels at pattern recognition but hallucinate under pressure. The missing half is symbolic AI, the world of mathematical logic, knowledge graphs, and hard constraints that provides precision, structure, and the guardrails enterprises actually need. Together, he argues, they map to what Dan Kahneman called fast and slow thinking — and just as humans need both, so does enterprise AI. From there, Ashwin addresses the trust gap holding back autonomous decision-making at scale, advising leaders to start with low-stakes decisions, build comfort gradually, and never hand real operational authority to AI in zero-tolerance environments — yet. He closes with a vivid picture of the autonomous enterprise five years out: AI handling all operational and execution-level decisions, humans moving up to tactical and strategic roles, and everyone shifting from problem-solving to problem specification. His parting message — everyone in the enterprise is about to get a double promotion, whether they're ready for it or not. If you liked what you heard today, please leave us a review - Apple or Spotify.  Learn more about your ad choices. Visit megaphone.fm/adchoices