Review

Automations: keyword rules

Use keyword-based automations to auto-rename and auto-type future uploads, and bulk-update existing rows.

Updated: 2026-03-03

Before you start

  • At least one reviewed transaction available
  • Entity type (personal or business) selected correctly

Open related features

Layered workflow

Layer 1: Capture a correction

  • Edit a transaction description in Review the same way you already do today.
  • When a keyword is not yet automated, the app offers a one-click “Create Automation” flow.

Layer 2: Define behavior once

  • Pick a keyword (for example: cursor), define the future description, and optionally set the category type.
  • Optionally apply the automation to all existing matching transactions immediately.

Layer 3: Maintain over time

  • Manage all rules in Automations page: edit, enable/disable, delete, and re-apply to existing data.
  • Future OCR and AI uploads both use active automations automatically.

What Automations does

Automations lets you save personal keyword rules so repeated merchant variations become consistent automatically. If your statements contain “Cursor subscription”, “Cursor AI subscription”, and “Cursor AI tool”, you can map keyword “cursor” to one standard description and type.

This removes repeated manual cleanup in Review and improves recurring detection quality because similar rows stay normalized over time.

Simple example

Suppose you edit one row description to “Cursor subscription”. The prompt appears and asks if you want to create an automation from that correction.

  • Keyword: cursor
  • Rename matches to: Cursor subscription
  • Set expense type: subscriptions
  • Apply to existing matches now: enabled

How matching works

Matching is case-insensitive and normalized, so “Cursor”, “CURSOR”, or “cursor ai tool” all match keyword “cursor”.

If multiple rules match the same row, the most specific keyword wins first (longer keyword), then the most recently updated rule.

Where to manage rules

Open the Automations page from the sidebar. You can view all rules by entity type, edit the keyword or target values, toggle active/inactive, delete rules, and run “Apply to existing” at any time.

Best practices

Follow these habits to keep automations safe and maintainable:

  • Use high-signal keywords (for example: cursor, netflix, notion) instead of vague terms (like pay or fee).
  • Keep one clear canonical description per merchant family.
  • Set category type only when you are confident the keyword is category-specific.
  • Review inactive rules quarterly and delete obsolete ones.
  • Always verify entity type before creating business vs personal automations.