In this episode of the Crazy Wisdom Podcast, host Stewart Alsop welcomes Leon Coe, founder of Amplify Intelligence, to explore the cutting edge of AI and its practical applications in workflows, agentic systems, and beyond. They discuss the evolution of autonomous agents, the shift toward combining traditional code with large language models, and how these integrations are reshaping both business processes and personal productivity. The conversation touches on the potential for AI to transform repetitive tasks, the role of probabilistic versus deterministic AI models, and Leon’s perspective on the future of a more AI-enabled economy. Leon also shares thoughts on balancing creativity with automation, especially for non-technical users, while Stewart probes the deeper implications of our increasingly AI-driven world. To learn more about Leon’s work, visit Amplify Intelligence or connect with him on Twitter @LeonJCoe. Check out this GPT we trained on the conversation! Timestamps 00:00 Introduction to the Crazy Wisdom Podcast 00:26 Understanding AI Agents 02:51 Agentic Workflows vs. Autonomous Agents 04:31 Productizing Large Language Models 06:56 Challenges and Innovations in AI Adoption 10:15 Enhancing Workflows with AI 18:30 Technical Insights and Practical Applications 25:29 Exploring Future Shock and AI Adoption 27:17 The Economic Impact of AI 29:01 The Evolution of AI Models 31:01 Voice AI and Its Potential 34:21 The Role of Social Media in AI Adoption 41:01 Historical Perspectives on Media and Misinformation 44:28 The Future of Media and AI 45:30 Navigating the Digital Age with AI 55:10 Concluding Thoughts and Future Discussions Key Insights Redefining AI Agents: Leon introduces a unique perspective on AI agents, emphasizing that they don’t have to be fully autonomous or self-directing to be useful. He defines an AI agent as simply the combination of large language models (LLMs) with traditional code, capable of producing workflows and taking actions. This broader view allows businesses to leverage agentic technology today by integrating LLMs into workflows in a controlled, deterministic way, similar to tools like Zapier but with enhanced intelligence.Adoption Challenges for Non-Technical Users: One of the biggest challenges in AI adoption, Leon argues, is bridging the gap between technical and non-technical users. Programmers and those familiar with building digital workflows can push AI’s capabilities further by crafting complex queries and structured workflows. However, non-technical users often lack this mental framework, so introducing them to AI involves not only teaching specific tools but also new ways of conceptualizing and using automation to make their lives easier.AI for Personal Productivity and Business Efficiency: Leon explains how AI can revolutionize business workflows by injecting intelligence at different points. For example, an AI-enhanced workflow could automatically extract key insights from meetings, turn transcripts into actionable summaries, and even create custom reports with minimal human intervention. This not only saves time but also minimizes the risk of tasks falling through the cracks, which can improve productivity across entire organizations.Cost Optimization and Model Selection in AI Workflows: In building effective AI applications, choosing the right model and managing costs are essential. Leon illustrates how selecting cheaper, high-performance models, like Gemini Flash instead of more expensive alternatives, can make workflows more efficient and cost-effective. By carefully balancing speed, token limits, and other parameters, companies can optimize their AI usage and control expenses without sacrificing performance.The Future of Databases and Dynamic Information: The conversation explores a potentially transformative shift in data storage and retrieval, where static databases might be replaced by API calls to AI models that generate information dynamically. This approach coul