A library author’s November 2026 update on web LLMs, with working Hashbrown examples, measured results, integration gotchas and recommendations.
Use a decision model to route input and pick UI, and a generative model to extract typed values. A real Angular and React example built with Hashbrown.
Hashbrown v0.8 adds typed decisions, UI selection, activities, web search, and token usage and finish reasons. Learn what is new and what changes when you upgrade.
atc is Hashbrown's new flagship example, a live map of Pacific Northwest air traffic with an assistant that answers in your own components. Here is how it is built, from the transponder to the pixel, in Angular and React.
Hashbrown v0.7 moves to AG-UI 1.0, adds createUiJsonSchema so agents Hashbrown doesn't run can answer with your UI kit, and makes core tree-shakable again.
Hashbrown v0.6 moves every connection between your app and your model onto AG-UI, adds interrupt and resume, shared agent state and server-executed tool calls, and ships a step-by-step migration guide with a prompt for your AI coding assistant.
Hashbrown v0.5 adds UI Kits, a rebuilt streaming JSON parser, improved Skillet schemas, Prose for React, and new docs for building generative UI in Angular and React.
Outline for the v0.4 release post—Prose upgrades, Chrome local models, new Bedrock and Anthropic adapters, Angular/React UI boosts, and fresh samples.
Hashbrown v0.3 introduces support for open-weight models via Ollama, MCP server integration, simplified UI schemas, new prompt helpers, and a redesigned docs site. Build generative UIs with free, open-source models.
We had 30 minutes to plan and 4 hours to build a playable, turn‑based Spotify playlist game using MCP with the help of Postman live on camera for the web dev challenge.
Hashbrown v0.2 delivers ergonomic tool-calling APIs, a stable JavaScript runtime with QuickJS, improved chat state management, and a complete React overhaul. Build generative UIs with better developer experience.
Introducing Hashbrown - a TypeScript framework for building generative user interfaces in Angular and React. Features streaming, tool calling, structured output, generative UI, and a JavaScript runtime for LLM-generated code execution.
Mike and Brian present Hashbrown at NG Rome 7th Edition as the closing keynote. Learn about building generative user interfaces in Angular with Large Language Models and the positive community feedback received.
Watch Ben, Brian, and Mike present Hashbrown on an episode of Angular Air. Learn about the framework for building generative user interfaces in Angular and React.
