Transform Node

Reshape records with map, aggregate, deduplicate, sort, and limit operations

Last updated 24 Jun 2026

The Transform node is the general-purpose data-reshaping step. It runs one of five operations on the incoming records: map (select, rename, compute, coerce types), aggregate (group and summarize), deduplicate, sort, or limit.

When to Use

Use Transform to normalize field names, coerce types, compute derived fields, roll rows up into group summaries, remove duplicates, reorder, or cap the row count before passing data downstream.

Configuration

The available fields depend on the selected operation.

Field	Type	Operation	Description
`operation`	`map` \| `aggregate` \| `deduplicate` \| `sort` \| `limit`	all	The transform to run (default: `map`)
`mappings`	array	map	Field operations, each `{ useExpression, source, expression, target, type }`
`includeUnmapped`	boolean	map	Keep fields not listed in `mappings` (default: `false`)
`onError`	`skip` \| `fail` \| `null`	map	What to do when a value fails coercion (default: `skip`)
`groupBy`	array	aggregate	Fields to group by, each `{ field }`
`aggregates`	array	aggregate	Aggregations, each `{ function, source, as }`
`keys`	array	deduplicate	Fields that define a duplicate, each `{ field }`
`strategy`	`first` \| `last`	deduplicate	Which duplicate row to keep (default: `first`)
`caseSensitive`	boolean	deduplicate	Treat key values case-sensitively (default: `true`)
`sortBy`	array	sort	Sort keys, each `{ field, direction, nulls }`
`limit`	number	limit	Maximum rows to keep (default: `100`)
`offset`	number	limit	Rows to skip before taking `limit` (default: `0`)

Map Mappings

Each mapping selects a source field and writes it to target. Set useExpression to compute the value with an expression instead of copying a field. Aggregation functions are count, sum, avg, min, max, count_distinct, first, and last.

Inline Expressions (map)

Map expressions interpolate field values with ${fieldName} and support safe arithmetic:

${first_name} ${last_name}     → concatenation
${revenue} * 0.1               → 10% of revenue

Example

Normalize a CRM export, keeping only mapped fields:

operation: map
mappings:
  - source: "Email Address"
    target: email
    type: string
  - source: "Annual Revenue"
    target: revenue
    type: number
  - useExpression: true
    expression: "${First Name} ${Last Name}"
    target: full_name
includeUnmapped: false
onError: skip

Notes

With includeUnmapped: false, only fields with an explicit mapping appear in the output.
type coercion options are auto, string, number, boolean, and date. On a coercion failure, onError decides whether to skip the row, fail the run, or substitute null.
In sort, each key chooses a direction (asc/desc) and null placement (first/last).
In aggregate, the output is one row per group, with one column per entry in aggregates.

Node Reference Complete reference for all automation node types, organized by category Score Node Compute a composite score from weighted criteria and optionally assign tiers Set Variable Node Compute named values from static input, fields, or expressions and store them on the automation Template Syntax Variable interpolation syntax for LLM, Agent, HTTP, and MCP node configurations

Still stuck? Browse the FAQ or contact us.