Building "Safe" Agents Using Claude Code: A Quick Tutorial

In the ever-accelerating pace of AI development, agents have now taken over as the “application layer” for AI in all industries. Really? Seriously ? Probably. There’s no question this is a critical avenue for workflow automation moving forward, but as always the tech is early (but promising), the hype is deafening, and trustworthy, actionable expertise is scarce. So we’re bringing one of the Bay Area’s most experienced product leaders, and agent builder, Michael Albers, to teach us how he built a serious, cool, and perhaps even safe agentic architecture, using multiple layers of trust. Over an incredible run in the tech industry Michael spent over 20 years in product leadership, most recently as CPO and SVP at Verizon Media (Yahoo! and AOL) - overseeing a $7.5B portfolio, 1B+ users, and a 200-person product organization. Today he runs Albers Advisory, where he works with growth-stage companies on product execution and AI-native workflows. He’s going to run us through how he built his agentic suite, step by step.