🚀 Getting Started

How Node Chains Work

Data flow between nodes: arrays, loops, parallel execution

Basic Principle

Each node in the chain receives data, processes it, and returns an array (list) of results. This array can contain one or many elements.

🎯 Key Rule:

If the previous node returned N elements → the next node will execute N times (once for each element).

Visual Example

Imagine this chain: RSS → GPT → Telegram

Step	Node	What Happens	Result
1	RSS	Reads the news feed	Returns 10 news items (array of 10 elements)
2	GPT	Rewrites each news item	Executes 10 times — once for each news item
3	Telegram	Publishes each result	Sends 10 messages

Every "pass" is completely independent. If GPT fails to process the 3rd news item, the other 9 will continue executing normally.

Conveyor Analogy

Imagine a conveyor belt with parcels:

Post Office (RSS) — brought 10 boxes
Inspection (GPT) — each box goes through the scanner individually
Delivery (Telegram) — each inspected box is delivered individually

Each box is a separate, independent cycle. They do not "know" about each other.

Exception: Batch Mode (Collection)

For some nodes, you can change the default behavior using the Execution Mode (execution_mode) setting. This is crucial for understanding how a node will process the received data list:

Individual (item) — default behavior. If the previous node returned 10 results, the next one executes 10 times (once for each element).
Batch (collection) — aggregation mode. The node receives the entire list of 10 elements at once and executes only one time.

💵 Savings and Speed: Using Batch mode not only speeds up the process but also saves significant credits. One node execution in aggregation mode is always cheaper than 10 separate runs of the same node.

For example, Google Sheet Append in Batch mode will add all 10 rows in a single request. It’s more reliable, faster, and cheaper than making 10 separate spreadsheet entries.

Test Run vs Scheduled Run

⚠️ Important Difference

There are two ways to run an automation:

Test Run (the "Run once" button on the canvas) — the "Only Unique" mode naturally is disabled. This is intentional so you always see the result and verify the node’s work.
Scheduled Run — all settings apply as configured. If "Only Unique" is enabled, previously seen data will be filtered out.

The same applies to the individual node test run — when you click the "Play" icon right on the node. In this case, the uniqueness mode is also ignored.

Demo

Automation chain example

← Back to Guide

🔗 How Node Chains Work