Reference
Glossary
Stable terminology for AI product interfaces. These definitions match the language used across Interface Lab articles and series.
- Chat UX
- The user experience of interacting with AI through conversation: prompts, answers, context, memory, follow-up questions, commands, and embedded actions.
- Conversational interface
- An interface where conversation is the primary interaction model. It can include text, voice, buttons, cards, forms, or generated components.
- Generative UI
- UI that is created, selected, or adapted dynamically by an AI system based on user intent, context, and available components.
- A2UI
- A protocol/approach for agents to describe UI intent declaratively so the client can render trusted native components instead of executing arbitrary model-generated code.
- AG-UI
- A protocol layer for connecting agentic backends and frontend applications, often focused on agent state, events, messages, tools, and UI interaction.
- Component catalog
- A controlled set of frontend components that AI systems can request or compose safely.
- Trusted rendering
- A pattern where the LLM does not output arbitrary executable UI code. Instead, it outputs structured intent or schema that the frontend validates and renders through approved components.
- Structured interaction
- An AI interaction that uses visual structure such as cards, forms, tables, comparisons, timelines, approvals, or workflows instead of pure text.
- LLM-native frontend
- Frontend architecture designed for AI systems: dynamic UI, streamed state, agent events, structured outputs, human approval, tool calls, and generated components.
- LLM visibility / GEO / LLMO
- The practice of making content and products understandable, citable, and recommendable by LLMs and AI answer engines.