ai-sdk

To use generateText and streamText with the AI SDK, you call these functions on the server and connect them to the useChat React hook for streaming outputs. This pattern enables real-time text generation and structured results in your React app.

Building a ToolLoopAgent with useChat involves defining tools and passing them to the agent function for safe tool invocation. The AI SDK handles the tool loop logic while streaming the intermediate and final text outputs directly to your chat interface.

Implementing a RAG system uses the embed function to generate vector representations of your documents via the AI Gateway. You then query these embeddings to retrieve relevant context and pass it to generateText for grounded AI responses.

Yes, the AI SDK supports multiple providers through the AI Gateway for provider selection. This allows you to switch between different language models seamlessly while maintaining consistent streaming outputs and structured results across your AI features.

You can use the useCompletion hook to handle streaming outputs and structured results from your model. It works alongside the AI SDK to manage completion states, ensuring safe tool invocation patterns and structured data handling in your UI.