Full Applications

Demo Apps

Self-contained applications that demonstrate A2UI concepts in realistic scenarios. Unlike experiments (which isolate primitives), demos show complete workflows with multiple agents, intent negotiation, and user governance.

Featured

Infrastructure Ops

Complete

Same data, different perspectives. AI surfaces what matters to each role.

AI-driven UIs should adapt to who is viewing, not just what data exists. This demo shows how Claude analyzes IT infrastructure data and proposes different UI elements based on role (DevOps Lead, SRE, Security Analyst).

role-based-uillm-integrationexplainabilityproduction-pattern

Store Canvas

Complete

Natural language queries become interactive visualizations for retail analytics.

Demonstrates Intent-Based Rendering with Composite Recursion using a retail store network domain. The LLM generates validated JSON intents that the frontend recursively renders.

intent-renderingcomposite-recursioncapability-negotiationdata-bindingmulti-proposal

shadcn/ui Analytics

Complete

A2UI intents rendered using production-ready shadcn/ui components.

Demonstrates how A2UI patterns integrate with existing component libraries. The same intent types are rendered using shadcn/ui Cards, Tables, and Recharts, proving the architecture is library-agnostic.

intent-renderingcomponent-libraryshadcnrechartslibrary-agnostic

Live Intent Playground

Complete

Edit A2UI intent JSON and watch it render in real-time with instant validation.

Understanding A2UI intents requires hands-on experimentation. This playground provides a Monaco editor with live preview, real-time validation, and pre-built templates to explore the intent contract.

intent-renderingdeveloper-toolslive-previewvalidation

Lab Analytics

Complete

This dashboard uses A2UI intents to render its own analytics - a meta demonstration.

The ultimate proof of A2UI is using it to build the tools that demonstrate it. This dashboard tracks page views and widget renders, displaying the data using the same IntentRenderer it measures.

intent-renderingmeta-demonstrationself-trackinglive-updates

UserAction Loop

Complete

Click buttons in rendered UI to trigger new LLM calls - demonstrating the full A2UI event loop.

A2UI enables continuous conversation through UI. This demo shows how user actions on rendered components trigger new LLM calls, creating an interactive loop without typing.

userActionevent-loopinteractive-uillm-integration

The Renderer

Complete

Understand how A2UI works: intents in, React components out. Interactive walkthrough of the rendering pipeline.

A2UI is a protocol, not a framework. The renderer is ~50 lines of code - a switch statement mapping intent types to React components. No special library needed.

rendererprotocol-vs-libraryframework-agnosticeducational

All Demos