In this episode, Carolyn welcomes back Dave Egts, Field CTO at MuleSoft, to explore the transformative power of AI in 2025. They dive into Dave’s top five AI predictions, with a special focus on the rise of agentic AI—AI that doesn’t just suggest, but acts. Dave explains the difference between generative AI and intelligent agents, why integration via APIs is the key to agent success, and how government agencies can prepare now.
What “large action models” (LAMs) are, and how they differ from large language modelsHow AI agents are already reducing friction in Salesforce and SlackWhy AI success depends on access to high-quality, actionable dataWhat’s driving the shift toward small language models and what that means for secure, on-device AIWhy “agent-washing” is the new “AI-washing”—and how to spot itWhat makes a truly impactful agent vs. a glorified chatbotDave’s favorite sci-fi robot, a monkey’s paw superpower wish, and what tech breakthrough he thinks is coming nextGuest: Dave Egts, Field CTO at MuleSoft, a Salesforce company
🔗 Follow Dave on LinkedIn
🎧 Listen to the Dave & Gunnar Show
[00:00] Intro & Dave’s blog: “5 AI Predictions for 2025”
[03:00] What’s an AI agent? Why they’re different from GenAI
[07:30] Real-world examples from Salesforce: Agentforce in action
[14:00] The importance of APIs for agent success
[15:45] Government use cases & security considerations
[18:00] Internal AI agents that reduce toil (meeting schedulers, IT help, expense reports)
[22:00] Cylon references & the power of digital labor in public sector
[29:00] Small Language Models: Why smaller might be better
[36:00] LLM Routers explained
[38:00] Fun Tech Talk Q&A (Superpowers, overhyped trends, sci-fi names)
📖 Dave's Blog: 5 AI Predictions for 2025: Opportunities Tech Leaders Must Seize This Year
💡 Trailhead Learning Platform from Salesforce: Learn to build your own agent
🤖 Help.Salesforce.com – Try Agentforce
📺 Forbidden Planet (Robby the Robot)
🐒 Monkey’s Paw story reference