Tambo
Generative UI SDK for React
Last verified:
What is Tambo?
Tambo is an open-source generative UI toolkit for React that enables product engineers to add AI-powered agents to their React applications. Instead of building static interfaces, Tambo lets AI agents decide which registered React components to render based on user messages. The agent renders your existing UI components with the right props, so
Tambo pricing
Pricing model: Freemium
Tambo offers a freemium pricing model with three tiers plus free self-hosting. Starter plan is Free with 10K messages/month, unlimited users (OAuth), chat-thread history, analytics + observability, and community support. Growth plan is $25/month with 200K messages/month included, $8 per additional 100K messages (billed in 100K blocks), unlimited users, chat-thread history, and analytics + observability. Enterprise plan is an Annual Contract with negotiated message volume, unlimited seats for cloud, enterprise-only features, SSO/SAML, SCIM, RBAC, single-tenant or on-prem deployment, 99.99% uptime SLA, SOC 2/HIPAA opt-in, GDPR upon request, early access to new features, and 24x7 support. The open-source version is self-hosted for free forever and includes the tambo.ai/react package, ui component library, and tambo-ai/tambo-cloud backend.
Tambo pros
- Open-source toolkit with free self-hosting forever
- Built-in AI agent requires no external frameworks
- Go from zero to agent in a weekend
- Uses your existing React components - register once, agent renders with right props
- Streaming support for progressive prop updates
- Automatic chat-thread history persistence
- MCP support for connecting to databases and APIs
- Auth inherits user permissions automatically for security
- Works with Next.js, Remix, Vite, and React Native
- BYOM (Bring Your Own Model) - flexible model selection
- CLI commands for quick setup: npx tambo init, npx tambo add, npx tambo full-send
- Two component patterns: generative and interactable components
- Error states, cancellation, and message threads handled out of the box
- Pre-built production-ready components via CLI installer
- Tambo Cloud REST API with OpenAPI specification for programmatic access
- Unlimited users on OAuth even on free Starter plan
- Active developer engagement and community on Discord and GitHub
Tambo cons
- Primarily focused on React ecosystem - not framework-agnostic
- Requires Zod schemas for component props which adds setup complexity
- TamboProvider only works in browser not server-side rendering
- .env.local file with API key cannot be committed to version control
- Free tier limited to 10K messages per month
- Growth plan charges $8 per additional 100K messages which can get expensive
- Enterprise features require annual contract negotiation
- Some documentation still marked as living document under active development
- Newer tool with less mature ecosystem compared to established AI frameworks
- Self-hosting requires managing tambo-cloud backend infrastructure
Frequently asked questions about Tambo
Is Tambo free to use?
Yes, Tambo has a free Starter plan with 10K messages per month, unlimited users via OAuth, chat-thread history, analytics, and community support. You can also self-host the entire toolkit for free forever, including the React package, UI component library, and tambo-cloud backend. Paid plans start at $25/month for the Growth tier with 200K messages.
How do I get started with Tambo?
There are two ways to get started: use the Tambo template project (recommended for new projects) by running npm create tambo-app my-app, or add Tambo to an existing Next.js/React application using npx tambo add or npx tambo init. The installation sets up Tambo in your project, gets you an API key, installs components hooked up to Tambo, and shows you how to wrap your app with TamboProvider.
What are generative components vs interactable components?
Generative components are rendered once in response to a user message - use these for charts, data visualizations, and summary cards that display results. Interactable components persist on the page and update by ID across conversations - use these for shopping carts, task boards, spreadsheets, and dashboards that users interact with over time. Interactable components register themselves automatically when mounted.
How does authentication work in Tambo?
For apps with signed-in users, you pass a userToken to TamboProvider which enables per-user authentication. The agent inherits your user's existing permissions, keeping your AI features secure without requiring separate auth implementation. This means your AI assistant respects the same access controls as your regular application.
What is MCP support in Tambo?
MCP (Model Context Protocol) support lets Tambo connect to databases, APIs, and external systems. You can connect pre-built MCP integrations or your own MCP servers, enabling the AI agent to fetch context and data from external sources similar to OpenAI tool calls. This allows Tambo to respond with real data from your systems rather than just static content.
Can I use my own AI model with Tambo?
Yes, Tambo is BYOM (Bring Your Own Model). You provide your own AI model API key (like OpenAI), and Tambo uses it for the agent capabilities. This gives you flexibility to choose your preferred model provider rather than being locked into a specific AI service.
What components are available out of the box?
Tambo provides pre-built production-ready components you can install via CLI using npx tambo add <componentname>. The component library includes components for forms, charts, messaging, and more. The MessageThreadCollapsible component provides a complete chat interface for your AI assistant. All components are installed into your /components/ui directory where you can customize behavior and styles.
How do I self-host Tambo?
You can self-host Tambo for free forever by using the open-source components: the @tambo-ai/react package for the frontend SDK, the UI component library, and the tambo-ai/tambo-cloud backend. This gives you full control over your infrastructure without relying on Tambo Cloud, though you'll need to manage the backend deployment and maintenance yourself.