Ep. 246: Deon Nicholas | A Glimpse of the AI Future—It's Here Today

Episode 246: The AI future of customer service is already here—and it's better than most people think. In this episode, Deon Nicholas, President and Executive Chairman of Forethought, joins host Rob Markey to show us how some companies are already using AI to resolve customer issues end-to-end in ways we could barely imagine just a couple of years ago. Deon introduces us to agentic AI: an emerging class of intelligent agents that take real action, integrate across enterprise systems, and adapt to each customer's needs. Drawing on his experience building Forethought's platform, Deon reveals how these systems are resolving issues, improving customer satisfaction, and going live in as little as 1 to 30 days. This isn't a future promise. It's happening now. The episode explores the architecture behind agentic AI, including Forethought's use of multi-agent systems, plain-language Autoflows, and a Discover model that learns company policies from historical tickets and call logs. Rob and Deon dig into risk, hallucination, and data privacy concerns—and how to address them without six-month implementation timelines. A surprising insight? Forethought sometimes adds a delay to its lightning-fast responses. Why? To build trust through operational transparency. Deon explains how even loading dots can reassure customers that the system is working on their behalf. Guest: Deon Nicholas, Founder, President, and Executive Chairman, Forethought Host: Rob Markey, Partner, Bain & Company Give Us Feedback: Help us improve the podcast (feedback link) Want to get in touch? Send a note to host Rob Markey. Key Topics Covered [1:00] Agentic AI vs. traditional chatbots [2:00] Why chatbots fail regarding decision trees and limitations [4:00] Real-time AI issue resolution and automation [7:00] AI integration with enterprise systems [12:00] Fast deployment and autoflow policy learning [15:00] Multi-agent AI systems and scalability [18:00] AI adoption challenges and business integration [22:00] Balancing AI automation and human agent handoffs [27:00] Cost comparison of AI vs. business process outsourcing [30:00] Rapid AI deployment and testing strategies Time-stamped Notable Quotes [4:00]  "With an agentic AI, you have something that has learned from your business policies. It's read through hundreds of thousands of past conversations, knows the vernacular, knows how to respond, and knows the business policy. So, instead of getting a menu of items, you just have a conversation." [13:00]  "You probably already have hundreds of thousands of conversations, whether they're sitting in transcripts, support tickets, [or] call recordings. It turns out that is a wealth of data that can train an AI in such a way that you don't need to manually create all these rules and decision trees and workflows." [ 16:00] "When we first launched our LLM-native AI two years ago, there were some hallucinations. But we've been able to go through, evaluate the model, fine-tune the model, and now we're at the point where it rarely happens. What we typically say to everyone is: 'Test it. Test it before you launch it, run a 14-day free trial, proof of value, run us against anyone else in the market.'" [21:00]  "What's beautiful about all of this is now you get to the point where AI can become embedded into the ecosystem—and, ironically, make all of these human experiences better." [21:00] "AI is making it so that when it's time to actually hand off to a human agent, you're far less frustrated, or far less likely to be frustrated. And then the humans will now be resolving issues that require more judgment and more empathy."