.webp)
Finance has its eyes on the AI bill. The problem, however, is seeing the full AI bill.
The invoice from the model providers, the direct token spend, the number with a name on it. That's the part they can see, and it's the part they're working hard to get their arms around.
It's also just one of the places the money goes. Underneath the invoice, in the cold and the dark, sits everything else. The hyperscaler running a model you forgot you deployed. The gateway routing your requests through a dozen providers and charging for the privilege. The coding tool billing by a unit you've never had to think about before. Finance isn't ignoring this spend. They can't see it in the first place.
That's the iceberg. And today, BRM reaches deeper into it than ever.
What's new: direct, API-level usage integrations
We've gone straight to the source. BRM now connects to the providers' own usage and cost APIs, reading spend where it's generated rather than waiting for it to surface on a statement weeks later. And we've done it across almost every layer of AI spend, each one harder to see than the last:
- Direct model providers. Anthropic, OpenAI, xAI. The token spend finance already watches, now pulled in automatically at the source.
- Hyperscaler-hosted models. Like AWS Bedrock, Azure AI Foundry, Google's Gemini Enterprise Agent Platform, Cloudflare Workers AI. The models running inside your cloud bill, where AI spend dissolves into a much larger number, and line items rarely say "AI."
- Gateways and aggregators. Like OpenRouter. One API key blending a dozen providers' rates, with routing fees stacked on top.
- AI-native applications. Like Cursor. Where the spend lives inside the product, and the model gets chosen by whoever's using it.
Each layer is harder to find than the one above it, and each is a place most tools never reach. A card statement stops at the vendor name. An observability tool lives in traces the finance team never opens. The integrations are how we read the usage. What turns that usage into intelligence is BRM's graph of your commercial relationships and the commercial models behind each line item: every charge resolves to the party it came from and the agreement that governs it, so raw consumption inherits the commercial model behind it. Spend scattered across five bills and a dozen consoles becomes one relationship you can see, own, and forecast.
Why it matters
You can't budget for what you can't see, and you can't manage what you can't name. With these integrations live, the questions finance has been forced to guess at start to have real answers. Questions like: What is this actually costing us? Which of it serves customers and which serves the team? What happens to the number at 2x the volume? The questions that decide a gross margin and a board slide, can finally become clear.
The AI meter has been running the whole time, in more places than most teams ever realized. The difference now is that you can finally read it.
Want to map your AI spend?
See your own AI spend, surfaced at the source.
More articles
Keep reading on procurement and strategy
Products
Introducing the Relationship
Every agreement, payment, conversation, and signal in one living record — introducing the Relationship View, the complete picture of every vendor relationship.
Best Practices
AI Token Spend 101: A Finance Leader's Guide to Understanding (and Managing) Usage-Based AI Costs
A finance-first guide to managing AI token spend: what tokens cost, why usage-based pricing is unpredictable, and a 6-step framework to baseline and forecast.
