Session abstract
What you’ll learn
What happens when software stops waiting for input and starts doing the work? We’ve spent decades building apps that sit still until someone clicks a button, fills out a form, or types a prompt. But as AI evolves, the real shift isn’t just in how we interact with software, it’s in what we expect it to do. We’re moving from app-first to agent-first design: from products organized around interfaces to systems organized around autonomous agents that can understand goals, pick up context, make decisions, coordinate with one another, and carry meaningful work forward. This isn’t just chatbot UX or automation with a new coat of paint. The future isn’t richer interfaces for every action. It’s systems that can act on their own, with interfaces reserved for delegation, visibility, and review, and sometimes no interface at all. Whether that means triaging inbound requests, coordinating research, or running a multi-step workflow, the goal isn’t to build better buttons. It’s to build software we can actually employ. In this talk, we’ll break down the core building blocks of agentic systems, including planning, memory, tool use, orchestration, UX, and oversight, and explore the architectural patterns that make them work. We’ll also look at what makes these systems trustworthy: guardrails, human approval where it matters, and clear visibility into what agents are doing and why. The result is a fundamentally different way to think about software: not as passive tools we operate, but as proactive coworkers we collaborate with and delegate to.