## Core Concept "Agentic apps will beat UIs. Users describe intent, agent does it." The interface paradigm shifts from designing screens users navigate to building agents that interpret and execute user intent. ## Key Insights - **Intent over interaction**: Users express what they want, not how to get it - **UI as legacy pattern**: Traditional interface design becomes intermediate step, not end state - **Conversation as interface**: Natural language becomes the primary interaction mode - **Complexity hidden**: Agent handles multi-step workflows user would previously navigate manually ## Design Implications - **From screens to capabilities**: Product design shifts from "what screens do we need" to "what can users accomplish" - **Error handling changes**: Agents must explain failures and offer alternatives, not just show error states - **Personalization at runtime**: Agent adapts execution based on user context, not preconfigured settings - **Trust as UX**: Users must trust agent actions they don't directly control ## Related Concepts - [[Voice as Last Interface]] - Brian Roemmele's prediction of voice as final interface convergence - [[AI Concept Externalization]] - Agent maintaining context externally - [[R×S Fit Checking Framework]] - Agent tracking requirements and solutions ## Cross-Domain Applications - **Enterprise software**: Complex workflows become conversational requests - **Consumer apps**: Feature discovery eliminated—users just ask - **Developer tools**: Claude Code as exemplar of agentic development interface - **Customer service**: Intent-based resolution replacing menu navigation ## Source - [[thoughts after a month of daily claude code]] - Brian Flynn, January 2026 --- *Atomic concept extracted January 2026*