Action

Batch

Gate each record, process passers

When one submission arrives holding many separate records, Batch splits them apart and screens each record on its own so only the ones that qualify continue on to the costly steps.

How it works

Sometimes a single job comes in carrying a whole stack of records at once: a spreadsheet of applicants, a folder of statements for several businesses, a batch of orders. Batch is the node that breaks that stack into its individual pieces and then runs the rest of your workflow once per piece, so one incoming job quietly becomes many.

As it splits, Batch also screens. You tell it what a good record looks like, and it lets the ones that pass move forward while holding back the ones that do not. That screen matters because the steps after it (reading documents, running analysis) usually cost credits per record, so filtering junk here keeps you from paying to process records you were going to throw away anyway.

The result is one clean fan-out point. Instead of a person opening a bundle and forwarding items one at a time, Batch does the separating and the first-pass sorting in a single step, and every downstream node then treats each survivor as its own little job.

Common use cases

Real situations where this node earns its place in a flow.

Process a spreadsheet of applicants

Gate each row against a rule and run the full pipeline only on the ones that pass.

Fan one file into many jobs

Expand a manifest into a job per record and process each independently.

Skip the ones that do not qualify

Run a batch where only passing records incur the expensive steps.

Batch in a workflow

A simple example of where this node sits in a flow.

CSV parser Spreadsheet → rows Batch Gate each record, process passers Extract PDFs → structured data Analyze Underwriting pass

What you set up

  • Point Batch at the incoming collection (the list or bundle of records) it should split apart.
  • Set the rule that decides which records pass the screen and which get held back.
  • Confirm which downstream steps should run once per passing record.

Common questions

What happens to records that do not pass the screen?

They are held back and never reach the expensive downstream steps, so you are not billed to process them. Only the passers fan out and continue.

Do I need Batch if my file already has one record?

No. Batch earns its place when a single submission carries many records that should each be handled on their own. For a one-record job it adds nothing.