Agentic UX

Agentic UX is the design of product experiences where agents, humans, and mixed human-agent workflows can all discover capabilities, inspect context, provide feedback, and complete tasks without being trapped in a blank chat box [src-088].

Key points

  • Mike Christensen's AI Engineer talk argues that broken AI UX is often not the model's fault; products fail when they hide capability, context, and state behind generic chat interactions [src-088].
  • RL Nabors's WebMCP talk adds a web-platform version of the same idea: sites should expose structured affordances to agents while retaining rich visual interfaces for humans [src-088].
  • Agentic UX needs visible state, recoverability, progressive disclosure, and evidence surfaces so users can understand what an agent did and intervene at the right moment [src-088].
  • The pattern connects Generative Ui, Webmcp, and Agent Native Infrastructure: agent interfaces should be interactive, inspectable, and tool-aware rather than only conversational [src-088].
  • Google's A2UI + MCP Apps patterns add an implementation layer: standard UI can be rendered through host-native A2UI components, while custom sandboxed interactions can live in MCP Apps [src-115].
  • [src-191] adds a product-facing version of the same lesson: agent output is not UX, prompts are still too punch-card-like, and AI products need visible state, explanations, receipts, rendered workflow surfaces, and user intervention points.
  • The playlist also frames AI product design as discernment rather than approval: good agentic UX should help users judge, correct, and trust work instead of nudging them to accept whatever the model proposes [src-191].

Related concepts

Source references

  • [src-088] AI Engineer late-May 2026 channel update (48 transcripts, 2026-05-15 to 2026-05-31)
  • [src-115] Google Developers Blog – "A2UI + MCP Apps" (2026-06-17)
  • [src-191] AI Engineer World's Fair Online Track 2026 playlist update (47 new transcript captures, 2026-06-22 to 2026-07-02)

Robin Cartier perspective

This page is part of Robin Cartier's working AI knowledge graph: a practical research layer for production AI, recommendation systems, experimentation, GEO, and agentic web readiness.

The useful next step is to connect this concept back to applied product leadership and operating models.

Recommended next

Keep reading from this thread

From 477 indexed pages and articles.

  1. Wiki concept MCP Apps Custom, sandboxed application surfaces that can be exposed through the Model Context Protocol and embedded into agentic user experiences when declarative Related by a2ui
  2. Wiki concept WebMCP A browser-side pattern that exposes page capabilities as agent-callable tools, making an HTML page behave like a small tool server for browser-native agents Related by webmcp
  3. Insight Agentic Web Readiness A practical checklist for making websites understandable, navigable, and useful for AI agents and answer engines Related by humans