Tag
The paper proposes an Inverse Theory of Mind (IToM) pipeline that infers user beliefs, preferences, and decision-making traits from observed interactions, using LLM-driven counterfactual reasoning to synthesize structured user personas for adaptive content recommendation across modalities, including a VisionOS spatial banking app.
Diyi Yang announces a talk with Yanzhe Zhang about generative UI at ACL2026, discussing new interfaces for human-AI interaction, with Zhang presenting their GenUI work in a poster session.
AG-UI is a protocol by CopilotKit for building agentic applications that stream lifecycle events, tool calls, and state updates, enabling human-in-the-loop workflows across multiple frameworks. The CopilotKit SDK implements this protocol.
An open-source tool inspired by Claude Tag, allowing users to bring their own model and runtime, with support for generative UI, streaming replies, and human-in-the-loop approvals.
Svpino criticizes Claude Tag as a vendor lock-in trap, while Atai Barkai announces Open Tag, an open-source alternative that works with any model and agent harness, supporting generative UI, streaming, and human-in-the-loop approvals.
The author proposes that traditional UI will be replaced by generative UI, comparing the transformation directions from six aspects: interaction pattern, state management, output method, task mode, observability, and security mode, showcasing new paradigms such as Agent-driven event streams and dynamic component trees.
CopilotKit, a framework for building generative UI and agentic frontends with LLMs, has been added to the Awesome LLM Apps repository (114K stars) in the new Generative UI and Agentic Frontends section.
MindLab releases Macaron-V1-Preview-749B, a Mixture-of-LoRA personal agent model built on a 744B base with 5 specialist LoRAs, featuring generative UI as a core skill and 202K context, under MIT license.
A tweet asserting that all UI will eventually be AI-driven, and that AI agents need generative UI to express information rather than generating plain text paragraphs.
Discusses three build patterns for generative UI (controlled, declarative, open-ended) and provides a decision tree for choosing the right approach based on use case.
The article explains three patterns of Generative UI (Controlled, Declarative, Open-ended) and how CopilotKit implements them via the AG-UI protocol for building dynamic agent-driven interfaces.
Google's announcements at I/O 2026 signal a massive shift in search from blue links to generative UI and personal AI agents, raising questions about how advertising will be integrated into LLM-generated output through methods like token auctions and prominence allocation.
Presents Macaron-A2UI, a model for generative UI in personal agents that synthesizes dynamic interfaces with lightweight executable actions, moving beyond text-only chat. The paper introduces a large-scale corpus, the A2UI-Bench benchmark, and trains models up to 754B parameters using LoRA fine-tuning and reinforcement learning, achieving strong results.
Improved lighting and materials on a coral reef demo using Three.js, with procedural caustics and emissive maps, and generated a UI using Omma AI.
DeepLearningAI released a free course on building Generative UI similar to Anthropic's Claude Artifacts, leveraging the open-source CopilotKit framework.
Google and Apple are bringing AI-powered 'vibe coding' to mobile, allowing users to create custom Android apps, widgets, and shortcuts via natural language prompts, as demonstrated at Google I/O 2026 and reported for iOS.
Google is transforming Search into an interactive execution layer that can generate interfaces and tools on the fly using Gemini 3.5 Flash and Antigravity framework, marking a significant shift in how users interact with information online.
Google at its I/O 2026 keynote announced new agentic AI features, Gemini 3.5 model updates, an intelligent search box with generative UI, and Android-powered smart glasses, aiming to embed AI agents across its services.
Google unveiled the biggest changes to Search at I/O 2026, including a redesigned search box with AI Overviews and AI Mode powered by Gemini 3.5 Flash, AI agents for Pro/Ultra subscribers, and generative UI for building apps in Search.
OpenUI is an open-source framework for generative UI that uses a streaming-first language to produce UI with up to 67% fewer tokens than JSON, including a React runtime and chat interfaces.