assistant-ui
Typescript/React Library for AI Chat💬🚀
Last verified:
What is assistant-ui?
assistant-ui is an open-source TypeScript/React toolkit that provides production-ready chat UI components and state management for building ChatGPT-style conversational interfaces quickly. It offers drop-in chat layouts with theming, streaming support, and sensible defaults so developers can focus on agent logic rather than UI plumbing. The project includes utilities for handling streaming, interruptions, retries, message editing, and multi-turn conversation state, and it integrates with major LLM providers and agent frameworks so it fits both simple GPT-wrappers and complex multi-agent apps. assistant-ui targets frontend engineers and product teams who want a polished, performant chat UX (web and React-based apps) with optional managed backend features for session persistence and developer-friendly integrations.
assistant-ui pricing
Pricing model: Freemium
assistant-ui core is forever free and open source under the MIT License, including the customizable UI components and the ability to bring your own backend; the website advertises a free managed ‘assistant-cloud’ tier to get started, while additional managed backend features, enterprise support, or Sales-contact options are available as paid offerings (details and limits for paid tiers are available via Contact Sales or the pricing page).
assistant-ui pros
- Open-source MIT-licensed library suitable for commercial use
- Drop-in ChatGPT-like UI components that reduce frontend work
- Built-in streaming support for responsive token-by-token rendering
- State management for multi-turn conversations and retries
- Optimized rendering and minimal bundle size for high performance
- Works with any LLM provider (bring-your-own backend)
- First-class integrations with LangGraph, LangChain, and Vercel AI SDK
- Supports rich content: markdown, code highlighting, tables, and charts
- Message editing and re-generation UX built into components
- Examples and AI-accessible documentation to accelerate adoption
- Includes a managed Assistant Cloud option for session persistence
- CLI starter commands (npx assistant-ui create / init) for fast setup
- Extensible AI tools and MCP support for IDE/assistant integration
- Used and tested by known projects and companies (ecosystem signals)
- Customizable theming and sensible default styles for production apps
- Human-in-the-loop patterns and tool UI patterns available
- Large community and GitHub releases with ongoing maintenance
assistant-ui cons
- Primarily React/TypeScript — not a drop-in for non-React stacks
- Backend responsibilities (APIs, auth, model billing) left to developer
- Some advanced agent integrations require additional configuration
- Managed cloud features are separate and may have usage limits
- UI opinionated defaults may require customization for niche designs
- Requires familiarity with streaming and agent concepts to use fully
- Documentation depth varies by topic; some advanced examples needed
- Not a full turnkey AI backend — you still wire LLM providers
- Potential bundle-size impact if not tree-shaken carefully
- Real-time collaboration or mobile-native SDKs are not provided
Frequently asked questions about assistant-ui
Is assistant-ui free to use in commercial projects?
Yes — the core assistant-ui library is open-source under the MIT License and available for commercial use; you can use the UI components and integrate with your own backend at no cost, while optional managed cloud services and enterprise support are offered separately.
Which frameworks and model providers does assistant-ui support?
assistant-ui is a React/TypeScript library that works with any LLM provider you wire into your backend and provides first-class integrations and examples for frameworks like the Vercel AI SDK, LangChain, LangGraph, and similar agent frameworks to simplify connection and stateful agent workflows.
Can I use assistant-ui with streaming token output?
Yes — assistant-ui is built for streaming output and optimized rendering so token-by-token or chunked model responses can be displayed responsively, including handling interruptions and retries within conversation state management.
Does assistant-ui store conversation history for sessions?
The core library itself is frontend-focused and leaves backend persistence to you, but assistant-ui offers a managed Assistant Cloud option that persists threads and session context so conversations can survive refreshes and context can accumulate over time.
How do I get started with a new project using assistant-ui?
You can bootstrap quickly using the provided CLI commands such as npx assistant-ui create for a new project or npx assistant-ui init for an existing project; documentation and examples on the website walk through installation, component APIs, and common integration patterns.
What rich content types are supported in messages?
assistant-ui supports rendering rich content including markdown, syntax-highlighted code blocks, tables, and charts, enabling developers to present structured outputs and developer-friendly formatting inside chat messages.
Is assistant-ui suitable for building multi-agent or tool-enabled assistants?
Yes — assistant-ui is explicitly designed to support complex agent patterns and tool UIs; it integrates with agent frameworks and includes examples and tooling for building multi-agent flows, tools, and human-in-the-loop interactions.
What are the customization options for appearance and layout?
The library exposes theming and styling hooks plus sensible defaults so you can customize colors, layout, and component behavior to match your product while leveraging the provided ChatGPT-like UX as a starting point.
Do I need to manage model billing and provider credentials?
Yes — when you bring your own backend or LLM provider you are responsible for provider credentials, API keys, and billing; assistant-ui focuses on frontend components and optionally offers a managed backend if you prefer not to self-manage persistence.
Where can I find examples and developer documentation?
Comprehensive docs, code examples, and a full plain-text documentation dump for AI assistants are available on the assistant-ui website and GitHub repo, including step-by-step guides, integration examples, and an index of component APIs to accelerate development.