TensorifyDocs

Core Concepts

Understanding these fundamental concepts will help you get the most out of Tensorify.

Workflows

A workflow is a directed graph that defines how data flows through your pipeline. Each workflow consists of:

  • Nodes - Individual processing steps (plugins)
  • Edges - Connections that pass data between nodes
  • Variables - Named data that flows through the graph

Workflows are visual representations of executable code. When you export, Tensorify compiles your workflow into clean Python.

Nodes

Nodes are the building blocks of workflows. Each node represents a plugin that performs a specific operation.

Node Types

| Type | Description | Example | | -------------- | ------------------------- | -------------------- | | Trigger | Starts workflow execution | Webhook, Scheduler | | Action | Performs an operation | OpenAI, HTTP Request | | Transform | Modifies data | JSON Transform, Loop | | Terminator | Ends a flow path | HTTP Response |

Node Anatomy

Every node has:

  • Settings - Configuration options (model, prompt, etc.)
  • Input Handles - Where data enters (left side)
  • Output Handles - Where data exits (right side)

Edges

Edges are the connections between nodes. They define:

  1. Execution order - Which node runs first
  2. Data flow - What data passes between nodes

Creating Edges

Click and drag from an output handle to an input handle to create an edge. The edge color indicates the data type being passed.

Variables

Variables are named pieces of data that flow through your workflow.

How Variables Work

  1. A node's output handle emits a variable
  2. The variable name is configured in the node settings
  3. Connected downstream nodes can access that variable

Example

[OpenAI] ---> response_text ---> [File Writer]

The OpenAI node emits response_text, which the File Writer receives and saves.

Execution Modes

Tensorify supports two execution modes:

Script Mode

  • Runs once from start to finish
  • Suitable for batch processing, data pipelines
  • Generated as a simple main.py

Server Mode

  • Runs as a persistent API server
  • Responds to HTTP requests
  • Generated as a FastAPI application
  • Triggered by Webhook or API Server plugins

Continue learning: