Composer

Generate typed, validated JSON data using Zod schemas and the generateObject function.

Render dynamic React components from AI tool calls using streamText and useChat.

AI generates structured JSON data that's automatically rendered as interactive visual components — cards, charts, tables, and more.

Basic non-streaming text generation with generateText. Submit a topic and get a generated paragraph — the simplest AI SDK pattern.

Generate images from text prompts using generateImage. Enter a description, click generate, and see the AI-created image.

Stream structured JSON output in real-time using streamText with Output.object and a Zod schema. Watch a recipe build up field-by-field as the AI generates it.

AI-powered code generation with syntax highlighting and copy support. Describe what you need, and the AI generates clean, runnable code.

AI generates dynamic forms from natural language descriptions using generateObject with Zod schemas. Describe a form in plain English and get a fully rendered, interactive form.

AI-assisted tabular data editing. Paste CSV data, ask AI to transform, analyze, or clean it, and see the results in an editable table with AI suggestions.

Stream text responses from AI models in real-time using streamText and the useChat hook.

A polished chat interface with rich markdown rendering — code blocks with syntax highlighting, tables, lists, headings, and inline formatting.

Show the AI's chain-of-thought reasoning in a collapsible thinking section, then display the final answer — like Claude's extended thinking.

A chat interface that includes inline source citations with expandable source cards. Responses include numbered references that link to source details.

Chat with images, files, and text in a single conversation. Drag-and-drop or paste images for vision analysis, attach files for context, and get AI responses that understand all modalities.

Let the AI model call functions and use external tools with the multi-step tool loop.

Build an agent that orchestrates multiple tool calls in sequence, using the AI SDK's maxSteps and automatic tool result forwarding.

Routes user queries to specialized sub-agents based on content analysis. A classifier agent determines intent and delegates to the appropriate specialist.

A parent agent that decomposes complex tasks, delegates subtasks to specialized child agents via tools, and merges their results into a cohesive response.

Generate-evaluate-iterate loop that self-improves output quality. Generates a draft, evaluates it against criteria, and regenerates until the quality threshold is met.

Integrate third-party search APIs (Tavily, Exa) with the AI SDK to build a grounded search agent with source citations.

Retrieval-Augmented Generation: embed documents, store in a vector database, and retrieve context for grounded AI responses.

Connect to any Model Context Protocol server, discover tools dynamically, and let the AI use them in a chat interface. Includes a demo MCP server with weather and calculator tools.

Natural language to SQL: ask questions about data in plain English, the AI generates and executes SQL queries, and displays results in a table. Uses an in-memory SQLite database with sample data.

Pause AI execution to get human approval before executing sensitive tool calls. Uses tool confirmation flow with the AI SDK.

Chain multiple AI calls in sequence, passing the output of one step as input to the next. Implements a research-summarize-translate-format pipeline.

Run multiple AI calls concurrently with Promise.all and merge the results. Performs sentiment analysis, entity extraction, and summarization in parallel.

Stateful conversation agent with Workflow DevKit that persists messages across pauses/resumes, calls tools, and streams responses using DurableAgent.

AI generates content, pauses via workflow hook until human approves/rejects via webhook. On approval, sends/finalizes; on reject, refines with feedback.

Durable workflow that sleeps for 7 days, then runs AI summarization/RAG on new data and generates a report. Uses 'use workflow' and sleep().

AI refinement loop inside a durable workflow: generate → evaluate score → if low, refine and loop (max 3 times). Uses 'use step' for each phase.