Rivet is an open source visual AI programming environment designed to help developers build complex AI agents using a node-based interface and prompt chaining workflows. It provides a desktop application that allows users to visually construct and debug AI logic as interconnected graphs, making it easier to manage sophisticated interactions between language models and external tools. Rivet also includes a TypeScript library that enables these visual graphs to be executed and integrated directly into applications, bridging the gap between prototyping and production use. Rivet supports multiple large language model providers and integrates with services such as embeddings and transcription systems, allowing developers to create richer AI-powered features. Its architecture emphasizes composability, where different components like prompts, APIs, and data processing steps can be combined into reusable pipelines.
Features
- Visual node-based editor for building AI agent workflows
- Prompt chaining system for composing multi-step AI logic
- TypeScript library for executing and embedding graph logic
- Support for multiple LLM providers and AI services
- Integration with embeddings and vector databases
- Desktop application for debugging and testing AI pipelines