Rivet

For teams struggling to build AI agents programmatically, Rivet offers a visual programming environment that simplifies the process of designing, debugging, and collaborating on complex LLM prompt graphs. As an open-source tool, Rivet is built and used by researchers, and its GitHub repository is actively maintained, with the latest release available for download. With Rivet, users can create visual, node-based graph editors to iterate on prompt graphs, which can then be run directly in their application, allowing for seamless integration and deployment. The tool's visual environment, easy debugger, and remote executor make it easier for teams to work together on complex projects, and its YAML-based graphs enable version control and review using standard code review tools. One notable limitation of Rivet is its current lack of support for multi-user collaboration in real-time, which can hinder larger teams working on the same project simultaneously. Despite this, Rivet's unique features, such as its ability to visualize and build complex chains for production, and its remote debugging capabilities, make it a valuable tool for those working with AI agents and LLMs, and its open-source nature ensures that it will continue to evolve and improve over time, driven by the needs of its community of users.