AI rule examples
Ready-to-copy rule recipes for common bookkeeping workflows.
AI rule examples
Admin+Five recipes you can copy and adapt. Swap in your own vendors and thresholds; the structure stays the same. The overview has the full list of fields and operators.
Recipe 1: tag every SaaS subscription
You want recurring software charges grouped under one tag so they are easy to find at reporting time.
- Phase: after
- Any (at least one matches):
- Vendor name contains
slack - Vendor name contains
notion - Vendor name contains
figma - Is subscription equals
true
- Vendor name contains
- Action: add the tag
SaaS subscription
The org-level Auto-tag Subscriptions toggle (Settings then General) does a lighter version of this in one click. Use a rule instead when you want to decide which vendors qualify or apply your own tag name.
Recipe 2: flag Amazon charges as reimbursable
Your finance team needs to spot Amazon purchases that staff paid for and should get back.
- Phase: after
- Any (at least one matches):
- Vendor name contains
amazon - Merchant name contains
amazon - Email from domain is
amazon.com
- Vendor name contains
- Action: add the tag
Reimbursable
The domain is operator matches subdomains too, so an email from marketplace.amazon.com still counts.
Recipe 3: skip a noisy vendor's order confirmations
One vendor sends an order confirmation and then a real invoice. You want only the invoice.
- Phase: before (runs before the AI reads the file, so it costs nothing)
- All (every condition matches):
- Email from domain is
noreply.somevendor.com - Email subject contains
order confirmation
- Email from domain is
- Action: exclude the document
A document excluded in the before phase never reaches extraction, so each match saves a credit.
Recipe 4: tag high-value invoices for review
Any invoice over $10,000 should get a second look before it is approved.
- Phase: after
- All (every condition matches):
- Total greater than
10000 - Document kind equals
invoice
- Total greater than
- Action: add the tag
Review needed
When the Email Routing add-on is on, the same condition shape can forward the email to your finance team.
Recipe 5: clean up a messy vendor name
The AI reads Amazon Web Services, Inc. off the invoice, and you would rather store it as AWS.
- Phase: after
- All (every condition matches):
- Vendor name contains
amazon web services
- Vendor name contains
- Actions:
- set Vendor name to
AWS - add the tag
Cloud infrastructure
- set Vendor name to
The set-vendor-name action writes to the vendor name on invoices and to the merchant name on receipts, since those are the matching field on each document type.
Add a rule
Open Rules
In the sidebar, open Rules (admin or owner only).
Create it
Click the button to add a rule and give it a name you will recognize later.
Pick a phase
After is the default and fits most recipes here. Pick Before only when you need to exclude a document ahead of extraction.
Build the conditions
Add items to the All group, the Any group, or both. An empty group adds no constraint, so leaving both empty matches every document.
Add the actions
Add or remove tags, set the vendor name, mark it as a subscription, exclude it, pass a hint to the AI, or stop the later rules.
Save
The rule runs on new documents right away. Older documents are not re-checked until you reprocess them from the document detail view.
Related
AI rules overview
Full field and operator reference.
Tags
What rules add and remove.
Vendors
Where a corrected name shows up.
Ready to try this?
Inbox Ledger turns your inbox into clean accounting data. The free tier includes 10 credits, refilled every 30 days.
Start free