Rivet
An open-source visual AI agent builder for macOS, Windows and Linux, with a TypeScript runtime for running prompt graphs in your own application.
Rivet is a desktop visual programming environment for developers building AI agents and applications with complex LLM workflows. Its editor runs on macOS, Windows and Linux, and its TypeScript library executes the resulting graphs inside your own application. The project uses the MIT license.
The visual graph shows how prompts and agent logic connect, so teams can inspect a workflow while developing it. Rivet also supports remote debugging: developers can watch graph execution in a running application and examine what happens along the chain. That makes the editor useful beyond the initial prototype, when the same workflow needs to run within a production application.
Graphs are YAML files that teams can keep in their existing repository and review through their usual code review tools. The runtime lets application code call a graph, and lets the graph call back into application code. This gives developers a visual place to work on agent behavior while retaining access to their application's existing functions.
Rivet integrates with OpenAI GPT models and Anthropic Claude, plus AssemblyAI LeMUR and speech transcription for audio workflows. It also supports OpenAI Embeddings and Pinecone for vector data. The editor and graph runtime run in your environment; these integrations connect workflows to external model and data services.
