Tambo 1.0: A Generative UI Toolkit for React

Tambo functions as a full-stack solution, providing both a React SDK and a hosted backend that manages conversation threads, agent execution, and state management. This eliminates the need for developers to build their own agent infrastructure from the ground up. The core workflow involves developers registering their existing React components with Tambo using Zod schemas to define the props. An integrated AI agent then interprets user requests, selects the appropriate component, and streams the props to render it dynamically. This allows a user to ask for "sales by region" and receive an interactive `<Chart>` component, not just a text description. Tambo is designed to work with a variety of large language models, including those from OpenAI, Anthropic, Gemini, and Mistral. It also supports integrations with agent frameworks like LangChain, though they are not required for basic functionality. A key feature is "tool calling," which allows the AI agent to interact with other functions and APIs. This empowers the AI to fetch data from an internal API or trigger workflows in other systems, transforming it from a simple UI generator into an agent capable of performing tasks. The toolkit distinguishes between two types of components: "Generative" components like charts that render once, and "Interactable" components such as a shopping cart that persist and can be updated through the conversation. State changes made by the user in the UI are communicated back to the agent, allowing for continuous and context-aware interaction. For developers looking to get started quickly, Tambo offers a free tier for its hosted cloud service, as well as the option to self-host the entire backend via Docker. The project is open-source and provides templates to bootstrap new applications, including an AI chat interface and an analytics dashboard.