Rivet is a visual programming environment for building, testing, and debugging AI agent workflows. You create AI pipelines by connecting nodes in a graph — like Node-RED but for LLM applications.
Why Rivet Matters
Building AI agents in code means blind debugging: you cannot see intermediate outputs, token usage, or where the pipeline fails. Rivet visualizes every step.
What you get for free:
- Visual graph editor for AI pipelines
- Real-time execution visualization
- Support for OpenAI, Anthropic, and local models
- Subgraphs (reusable AI components)
- Built-in prompt testing and evaluation
- TypeScript SDK for production deployment
- Node types: LLM, text, code, conditions, loops
Quick Start
# Install the desktop app from rivet.ironcladapp.com
# Or use the TypeScript SDK
npm install @ironclad/rivet-node
Visual Graph Components
[User Input] → [Prompt Template] → [Claude/GPT] → [Extract JSON] → [Output]
↑ |
[System Prompt] [If Error] → [Retry with different prompt]
Each node is configurable:
- Chat Node: Call any LLM (GPT-4, Claude, Llama)
- Text Node: Template with variables
- Code Node: Run JavaScript for transformations
- If/Else Node: Conditional branching
- Loop Node: Iterate over arrays
- Subgraph Node: Reuse entire pipelines
TypeScript SDK (Production)
import { runGraph, loadProjectFromFile } from "@ironclad/rivet-node";
// Load your visual graph
const project = await loadProjectFromFile("my-agent.rivet-project");
// Run it programmatically
const result = await runGraph(project, {
graph: "Main",
inputs: {
userMessage: { type: "string", value: "Analyze this data..." },
},
context: {},
openAiKey: process.env.OPENAI_API_KEY,
});
console.log(result.output.value);
Node Types
| Node | Purpose |
|---|---|
| Chat | Call LLM (GPT, Claude, Llama) |
| Text | Template strings with variables |
| Prompt | System/user/assistant messages |
| Extract JSON | Parse structured output |
| Code | Run JavaScript |
| If/Else | Conditional logic |
| Loop | Iterate over arrays |
| Subgraph | Reuse pipelines |
| HTTP | Call external APIs |
| Vector Store | RAG retrieval |
Links
Building AI agent pipelines? Check out my developer tools on Apify or email spinov001@gmail.com for custom solutions.
Top comments (0)