Insights

How do I book OpenAI or Anthropic invoices in NetSuite?

By Brian Diamond

Published October 5, 2026

Book OpenAI and Anthropic invoices in NetSuite by entering each provider as a vendor, coding the Vendor Bill to an AI services expense account with Department, Class, and (if used) Location segments on each line, accruing metered usage at month end with an auto-reversing Journal Entry, and letting the bill clear the accrual so only the true-up lands in the new period. The provider invoice is a single line with no attribution; the segments on the bill have to come from your own metering.

Why the invoice alone is not enough

An OpenAI or Anthropic invoice shows total usage by model, sometimes by project or workspace, never by your departments or customers. Workspaces and API keys can be mapped to cost objects if you set them up that way (one key per agent is the simplest discipline). Otherwise attribution comes from a gateway or a metering system that saw every request.

Set up once

  1. Accounts. Expense "AI Services" (sub-accounts optional) and liability "Accrued AI Services."
  2. Segments. Department for the consuming business unit. Class for the agent or use case if you want per-agent reporting without new accounts. Customer or Project for billable work. A custom segment "AI Agent" is cleaner than overloading Class if you have many agents.
  3. Vendors. OpenAI, Anthropic, and any gateway or marketplace reseller, with payment terms matching their billing (typically net 30 from monthly invoice; marketplace purchases may flow through the cloud vendor instead).
  4. Memorized reversing journal. The month-end accrual as a memorized Journal Entry with "Reversal Date" set to the first of next month.
  5. Expense allocation schedule (optional). For a shared agent, an allocation schedule can split a single bill line across departments by percentage.

The monthly cycle

  • Month end. Post the accrual JE: AI Services by Department and Class, Accrued AI Services credit. It auto-reverses on the first.
  • Bill arrives. Enter the Vendor Bill to AI Services with the same segments, using the metering split (or the allocation schedule).
  • Net effect. Accrual in period one; reversal plus actual in period two leaves only the variance.

Worked example

September metered usage: OpenAI $5,000 (Marketing agent 60%, Support agent 40%), Anthropic $3,200 (Research agent 100%, Product department).

Journal Entry, September 30, reverse October 1

Account Debit Credit Department Class
AI Services 3,000 Marketing Marketing agent
AI Services 2,000 Support Support agent
AI Services 3,200 Product Research agent
Accrued AI Services 8,200

Vendor Bill, OpenAI, October 7, $5,075

Line Account Amount Department Class
1 AI Services 3,045 Marketing Marketing agent
2 AI Services 2,030 Support Support agent

Vendor Bill, Anthropic, October 9, $3,200

Line Account Amount Department Class
1 AI Services 3,200 Product Research agent

October carries a $75 true-up on OpenAI, split 60/40, and zero on Anthropic.

Prepaid credits and marketplace purchases

If you buy credits in advance, book the purchase to a Prepaid AI Services asset and amortize to expense by metered usage each month, by Department. If you buy through AWS, Azure, or Google marketplace, the charge arrives on the cloud bill; code that line to AI Services with the same segments rather than leaving it in the cloud infrastructure account.

Reports

Income Statement by Department shows AI Services per business unit. A saved search on AI Services grouped by Class gives cost per agent. Both read directly from the segments posted above, which is why the segments matter more than the accounts.

Related

How do I account for AI agent spend in the general ledger? · What is the journal entry for AI token usage?

Feature names reflect NetSuite as of this writing and vary by edition and configuration. Segment design should be agreed with your controller and NetSuite administrator.

Not sure which of your AI costs are being booked? Run the free Agent Spend Assessment.

Onaro Meridian is FinOps for agentic AI: the system of record that attributes, controls and books what AI agents spend.

Brian Diamond

Brian Diamond

Brian Diamond is a fractional Chief AI Officer and founder of Onaro. He has spent 30 years running infrastructure operations and founded LANStatus, a Connecticut managed services provider and Microsoft partner, in 2001. He holds a Chief AI Officer certification and writes the CAIO Brief on AI leadership for finance and operations.

LinkedIn · CAIO Brief · Author page

Markdown version