Action

Classify

Select Case · branch per rule

The Classify node sorts each job by your own rules and sends it down the right path. Think of it as the fork in the road: eligible work goes one way, anything that fails a rule goes another, all without a person checking each one.

How it works

You define the rules once, in plain terms (for example: reject if there are more than three overdrafts, or route business accounts and personal accounts separately). Classify checks every job against them and picks the matching path.

Each path is a separate output you wire to different next steps. The eligible path might continue to analysis and a report; a failing path might go straight to a decline notice. Nothing sits in a queue waiting for a human to triage it.

Because it runs on rules, not AI, it is fast and free to run, which makes it the ideal gatekeeper in front of the paid steps.

It also fails safe on a location rule it can't evaluate: if you gate on the applicant's state but the state couldn't be read from the application or the statements, the deal isn't quietly passed through — it's held for manual review before it can be submitted to a CRM, so a restricted-state deal with an unreadable address can't slip by unchecked.

Common use cases

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

Decline before you spend on AI

Screen out applications that fail a hard rule (too new, wrong state) so only viable ones reach the paid analysis step.

Send each case down its path

Branch a job by policy (high risk, standard, fast-track) and run a different tail for each.

Flag the ones needing a human

Classify jobs into buckets so the tricky ones get routed to review and the clear ones flow through.

Classify in a workflow

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

eligible reject Extract PDFs into structured data Classify Branch on your rules AI Analyze Eligible path Decline Reject path

What you set up

  • Write your rules as named cases (each becomes an output path on the node).
  • Connect each path to what should happen next for that case.
  • Jobs that match no case fall through a default path you control.

Common questions

How is this different from Analyze?

Classify follows fixed rules you write, so it is instant and free. Analyze uses AI to form a judgment. Classify is the cheap filter you put in front of the expensive step.

What happens to a job that fails a rule?

It leaves on the path you connected to that case, for example straight to a Decline node that sends a rejection notice. You decide what each path does.