Build generative UI for AI agents on Amazon Bedrock AgentCore with the AG-UI protocol
Official AWS Machine Learning Blog walkthrough for deploying AG-UI agent frontends on Amazon Bedrock AgentCore. It shows FAST patterns for Strands Agents and LangGraph, then layers in CopilotKit for inline generative UI, shared state, and human-in-the-loop flows with Cognito, Lambda, Amplify, and AgentCore Runtime.
Add MCP Apps to your AI SDK application
Vercel Knowledge Base guide for building an MCP Apps host with AI SDK. It shows how `@ai-sdk/mcp` and `@ai-sdk/react` split model-visible tools from app-only tools, read `ui://` resources, and render interactive tool UIs inside a sandboxed iframe with `experimental_MCPAppRenderer`.
ag-ui-protocol/ag-ui
The official AG-UI protocol repository, positioned as the agent-user interaction layer for bringing agents into frontend applications. It is especially relevant because it sits one level below app-specific demos: it defines the transport and event model that lets agent backends stream messages, tool activity, and UI updates into real products across clients.
A2UI + MCP Apps: Combining Declarative and Custom Agentic UIs
Official Google Developers Blog architecture post on combining A2UI with MCP Apps. It is useful because it moves beyond standalone protocol announcements and shows three concrete integration patterns: serving A2UI payloads over MCP, embedding MCP Apps inside A2UI components, and running A2UI renderers inside MCP Apps.
syntux - Build Generative UIs for the Web
Syntux is an open-source React library focused on generating web interfaces from a supplied value plus optional UI hints. It uses a flat JSON-DSL called React Interface Schema, supports streaming, custom components, caching, and reactive regeneration, making it a useful reference for data-to-interface GenUI rather than chat-only assistant surfaces.
AG-UI Integration with Agent Framework
Microsoft Learn documentation for connecting Agent Framework agents to AG-UI frontends. It shows how Microsoft frames AG-UI inside production agent apps, including agentic generative UI, tool-based UI, and the event bridge between the backend agent runtime and React clients.
A2UI v0.9: The New Standard for Portable, Framework-Agnostic Generative UI
Official Google Developers Blog update on A2UI v0.9. The post positions A2UI as a production-oriented standard for declarative, agent-generated UI, adds the official React renderer, and explains how agents can target existing component catalogs across web and mobile without generating arbitrary frontend code.
Stop making AI write JSON - Why we built OpenUI
OpenUI's official engineering write-up on why the team moved beyond JSON trees for Generative UI. It lays out the runtime requirements for interactive agent-generated interfaces, argues that nested JSON becomes fragile for stateful UI, and introduces OpenUI Lang as a streaming-first language for model-generated UI.
Introducing Tambo 1.0
Official Tambo launch post for its production-ready React generative UI toolkit. The post explains the fullstack model: developers register existing React components, the agent selects and streams props into them, and Tambo handles conversation state, orchestration, and persistent interactive component behavior.
MCP Apps - Bringing UI Capabilities To MCP Clients
Official Model Context Protocol announcement for MCP Apps, the first production-ready MCP extension for returning interactive UI directly inside AI clients. It explains how tools declare `_meta.ui.resourceUri`, how hosts render sandboxed interfaces, and why dashboards, forms, visualizations, and multi-step workflows fit better than pure text for many agent tasks.
Introducing A2UI: An open project for agent-driven interfaces
Official Google Developers Blog announcement for A2UI, a portable declarative format for agent-generated interfaces. It explains how remote or local agents can request trusted UI components from a host app instead of emitting arbitrary HTML, making Generative UI safer, cross-platform, and incrementally updateable.
Google Research Blog: Generative UI
Google Research blog article "Generative UI: A rich, custom, visual interactive user experience for any prompt". Written by senior engineers at Google Research, it introduces the implementation approach and experimental results of Generative UI. The article states: "Generative UI is a new capability that enables AI models to create not just content but the entire user experience", and explains that this feature has been experimentally launched in Gemini app and Google Search AI Mode.
Google launches Gemini 3, Google Antigravity, generative UI features
Constellation Research technical analysis article introducing the Gemini 3 launch and emphasizing Generative UI features: Gemini 3 can "create interactive components, provide different scenario options, images, tables and text". The article details the Visual Layout and Dynamic View experiments.
9to5Google: Gemini 3 Launch Analysis
9to5Google detailed analysis of Gemini 3 explaining Generative UI features. The article notes that Dynamic View enables Gemini 3 to "design and code a fully custom interactive response" for each prompt, while Visual Layout generates "magazine-style" immersive multimedia interfaces.
Generative UI Project - Google
Google team's Generative UI project website (generativeui.github.io), containing research papers, interactive examples, and some code, demonstrating their GenUI implementation approach.
AIBase News: Google Generative UI
Chinese tech media AIBase's coverage in November 2025 reporting that "Generative UI enables AI to generate actionable dynamic interfaces when answering questions".
Generative UI with AI: The Future of Frontend Web Development
Comprehensive article by Pansofic exploring how Generative UI with AI is transforming frontend development. It covers real-time UI adaptation, natural language to interface generation, personalized experiences, and faster prototyping. The article includes case studies of Builder.io, Vercel AI SDK, and Figma AI Assistant, along with best practices for implementation.
AI is the new UI: Generative UI with FastHTML
In-depth tutorial by Pol Alvarez Vecino demonstrating how to build interactive Generative UI applications using FastHTML and HTMX in less than 150 lines of code. The article explores the evolution from text-only chat interfaces to display-only GenUI and fully interactive GenUI, explaining how the hypermedia approach eliminates "contract coupling" between frontend and backend, enabling LLMs to generate truly dynamic interfaces.
Generative UI: Smart, intent-based, and AI-driven
Medium/Design Bootcamp article by Daniel Ostrovsky explaining the potential of Generative UI from business and user perspectives. The article argues that GenUI is not about AI "creating entirely new components from scratch", but rather intelligently selecting and arranging existing interface elements based on user intent. It demonstrates how GenUI can improve efficiency and experience through examples in financial services and educational applications.
An introduction to Generative UIs
UX Collective article by Mark O'Neill introducing the concept and practical cases of Generative UI. The author notes that this term "emerged in 2023" and defines it as "technology that automatically builds or adjusts interfaces based on context through generative AI". The article demonstrates how UI elements, layouts, and styles can vary by user to achieve personalized experiences with GenUI.
Generative UI and Outcome-Oriented Design
Article by Nielsen Norman Group discussing the impact of Generative UI on design paradigms. The author defines Generative UI as "technology that dynamically generates tailored interfaces for users in real-time via AI", arguing that it will shift interface design from satisfying the majority to satisfying individuals. The article advocates transitioning from traditional interface design to "outcome-oriented design", focusing on user goals and final experiences.
MCP-UI-Org/mcp-ui
The official MCP-UI SDK repository for building and rendering interactive UI over MCP. Its README is especially relevant because it documents the post-standard shape of the ecosystem: `@mcp-ui/*` now implements the MCP Apps standard, supports TypeScript, Python, and Ruby server paths, and remains a practical host/client SDK for teams shipping UI-rich agent tools.
CopilotKit Docs: Generative UI Overview
Official CopilotKit documentation page that explains the core primitives behind its Generative UI stack. It breaks the space into concrete runtime patterns such as components-as-tools, tool-call rendering, state rendering, sandboxed MCP Apps, and declarative UI renderers, making it a useful map of production implementation choices.
CopilotKit/generative-ui
Official CopilotKit example repository that demonstrates three concrete Generative UI patterns: controlled GenUI with frontend tools, declarative GenUI with A2UI/Open-JSON-UI, and open-ended UI generation rendered inside a sandbox. Compared with generic demos, it is valuable because it places multiple GenUI approaches side by side in one maintained example set.
CopilotKit/OpenGenerativeUI
CopilotKit's official open-source framework for open-ended Generative UI. Unlike declarative component-spec examples, this repo focuses on richer sandboxed HTML/SVG experiences such as charts, diagrams, simulations, and 3D scenes, showing how an agent can generate full interactive visual responses inside a controlled iframe runtime.
thesysdev/openui
The official OpenUI GitHub repository from Thesys. It packages the OpenUI language, runtime, and component-driven React integration in one place, making it the concrete implementation companion to the team’s design essays. It is especially useful for developers evaluating streaming-first GenUI systems beyond plain JSON schemas.
google/A2UI
The official A2UI repository from Google. It packages the specification, renderers, samples, and tooling for the agent-to-user-interface format, making it the most direct implementation reference for portable, declarative Generative UI across trusted component catalogs.
Flutter GenUI SDK
Official library (BSD license) from the Flutter team for integrating generative UI capabilities in Flutter applications. It provides JSON-formatted interactive component definitions and state feedback mechanisms, converting chat outputs to actionable interfaces. Currently has 930+ stars and is an experimental project from Flutter.
OpenAI Structured Outputs Sample: Generative UI
Official OpenAI sample app showing a practical structured-output path to Generative UI. The demo uses a recursive component schema plus a `generate_ui` tool so model responses can stream reliable UI definitions instead of only text, while still supporting multi-turn conversations and automatic tool execution.
vercel-labs/json-render
Vercel Labs' official open-source Generative UI framework. json-render lets models generate interfaces from prompts while staying constrained to a developer-defined component catalog, then progressively renders the resulting spec across web, mobile, and other runtimes.
Hashbrown - The TypeScript Framework for Generative UI
Hashbrown is an open-source framework for building browser-native agents and Generative UI in React and Angular. Its strongest fit for this collection is the component-control model: developers expose trusted UI components, while the LLM composes dynamic views, streams them into the page, calls client-side tools, and works with structured data instead of only text.
mdocUI - Generative UI with Markdoc Tags
mdocUI is an alpha Generative UI library that asks LLMs to write normal Markdown with inline Markdoc-style component tags. The project is still early, but it is useful because it represents a distinct approach from JSON trees and tool calls: prose and interactive components can share the same stream while the renderer handles validation, placeholders, error boundaries, and custom component mapping.
LangUI - UI for AI
LangUI (MIT license) is an open-source Tailwind CSS component library customized for GPT/generative AI applications. It provides rich UI components that can integrate with any LLM-powered project (such as chatbots, content generation tools), simplifying GenUI development.
Anilturaga/Generative-UI - Imagine with Claude
Open-source "Imagine with Claude" demonstration project, showcasing generative UI capabilities with Claude.