Rivet

Rivet is a desktop visual AI agent builder for developers working with connected prompts and application logic. Teams can inspect complex LLM workflows as graphs and run those graphs inside their own applications.

• Visual workflows: Build prompt chains and agent logic in a graph editor that shows how the parts connect.
• Remote debugging: Observe a graph executing within a running application and inspect what happens along the prompt chain.
• Application integration: The TypeScript runtime lets application code call Rivet graphs and lets graphs call back into the application's functions.
• Team review: Store graphs as YAML files in your repository, track changes and review them with your existing code review tools.
• Model and data services: Connect workflows to OpenAI GPT and Anthropic Claude, use AssemblyAI for transcription and LeMUR voice processing, or work with OpenAI Embeddings and Pinecone.

The editor runs on macOS, Windows and Linux. Its graph runtime runs in your own environment, including Node and TypeScript applications. The listed model, speech and vector integrations use external services, so those parts of a workflow send data outside the local editor and runtime.

Rivet is open source under the MIT license.