Rivet is a visual IDE and TypeScript library for creating complex AI agents and prompt chaining workflows.
Rivet addresses the challenge of building and orchestrating multi-step AI interactions by providing a node-based visual programming environment where developers can design agent workflows without writing code. The desktop application lets users connect LLM calls, embeddings, and other AI operations into directed graphs, then execute and test them interactively. The underlying TypeScript library, Rivet Core, allows these graphs to be embedded and executed within applications, enabling bidirectional integration where Rivet graphs can invoke application code and applications can invoke Rivet graphs.
Rivet suits teams building AI agents, retrieval-augmented generation systems, or complex prompt chains who want a visual workflow editor combined with programmatic control. The tool integrates with multiple LLM providers including OpenAI, Anthropic Claude, and AssemblyAI, and supports vector databases like Pinecone and OpenAI Embeddings. Developers can use the desktop application for design and testing, then deploy the same graphs via the TypeScript library in production environments. The project offers both a standalone application and a library-based approach, making it flexible for different deployment scenarios.
The project maintains active engagement with its community through a Discord server and GitHub discussions. Development activity shows consistent attention to feature requests and bug reports across multiple integration points. The codebase is organized to support both the visual application and the programmatic library, with separate NPM packages for core functionality and Node.js-specific features. Documentation is maintained alongside the application to guide both visual workflow creation and programmatic integration patterns.
