Salt AI Docs

Getting Started

Create your first Salt pipeline in minutes.

Salt lets you build AI-powered pipelines by connecting nodes on a visual canvas. This guide walks you through creating, running, and understanding your first pipeline.

What You'll Learn

By the end of this guide, you will have:

  • Created a new pipeline in the Salt editor
  • Added and configured nodes
  • Connected nodes to create a data pipeline
  • Run your pipeline and viewed the results

Prerequisites

You need a Salt account to get started. If you don't have one yet, sign up at salt.ai.

Create Your First Pipeline

Open the Pipeline Builder

After signing in, navigate to the Pipelines page. Click New Pipeline to open the visual pipeline builder with a blank canvas.

Add Your First Node

Press A to open the node browser. You can browse categories or use the search bar to find nodes by name.

Drag a node from the browser onto the canvas, or double-click to add it. Each node represents a single processing step -- for example, generating text with an LLM, transforming data, or calling an external service.

Configure the Node

Click a node to select it. The Properties panel opens automatically, where you can set the node's inputs using widgets like text fields, dropdowns, number sliders, and file pickers. You can also press P to toggle the Properties panel at any time.

Fill in the required inputs for your node. Required fields are marked and the node will indicate if configuration is incomplete.

Add a Second Node and Connect Them

Add another node to the canvas. To connect two nodes, drag from an output handle on the first node to a compatible input handle on the second node. Handles are color-coded by data type -- you can only connect handles that share a compatible type.

The edge creates a data flow: when the pipeline runs, the output of the first node feeds into the input of the second.

Run the Pipeline

Click the Play button in the floating toolbar at the bottom of the canvas. Salt sends your pipeline to the execution engine and you can watch progress in real time.

Each node highlights as it processes. When execution finishes, click on any node to inspect its output in the Properties panel.

Next Steps

Now that you have a working pipeline, explore these topics to go deeper:

On this page