Ramp launches AI token spend tool for finance teams

Ramp has launched AI Token Spend Management, a product designed to give finance teams visibility into spending on artificial intelligence models across multiple providers.

The tool brings together spending data from providers, including Anthropic, OpenAI, Gemini and Cursor, in a single dashboard. Users can sort spending by provider, model, user or API key, and organise it by team or project.

Ramp is targeting a fast-growing cost area that many finance departments struggle to track. AI spending often sits across provider dashboards, card transactions and invoices, while usage is commonly tied to API keys rather than clearly assigned budget owners.

Ramp said it developed the product with more than 1,300 businesses that manage more than 100 trillion tokens a month. It is also offering businesses free access to view total AI spending and where that money is going.

Finance visibility

The launch reflects a broader shift in how companies are trying to govern AI costs. Finance teams have long had systems for travel, expenses and software procurement, but token-based billing for AI services can change quickly and is harder to explain when usage rises unexpectedly.

The product is intended to help finance teams distinguish between spending increases linked to business growth and avoidable waste. It also aims to show whether a more expensive model is justified or whether a lower-cost option could deliver similar results.

One part of the product is a weekly briefing that explains major week-on-week changes in spending, flags unusual spikes and suggests ways to cut costs. Ramp said the briefing translates technical choices into terms that finance staff can use when discussing budgets with engineering teams.

"As CFO, I need to give stakeholders timely financial information so they can make informed decisions. Ramp's Token Spend gives us real-time visibility into AI spend across providers, by user, model, and team, without burdening finance with manual reporting," said Christy Schwartz, Chief Financial Officer at Opendoor.

Cost controls

The tool also lets companies set soft or hard spending limits by user or API key. It can send anomaly alerts through Slack or email and notify managers, key owners and other team members when spending patterns change.

Ramp said the system suggests limits based on observed trends. That is meant to offer more targeted controls than blanket restrictions that can slow teams down.

Ramp said it found potential savings equal to 12% of monthly AI spending for the average business using the service. It cited examples, including a $1,600 weekend spending spike tied to one user and a case in which a team spent more than $3,000 on Anthropic's Opus model for work that the lower-cost Sonnet model could have handled.

Another example involved a company where the use of Fast Mode increased costs sixfold in one week, even though the task did not require faster output. Ramp said the setting was then turned off.

"Ramp's Token Spend Management briefing surfaced 'prompt caching,' not something on my radar as a Controller. I routed it to engineering immediately and we found we'd been losing $10,000 a month. We implemented the fix the same day," said Greg C., Controller at AngelList.

Growing spend

The launch comes as businesses face rising AI adoption and shifting model pricing. Those changes have created pressure on finance leaders to understand not just how much they are spending, but also which teams are responsible and whether the spending matches business value.

By linking spending to users, teams, and projects, Ramp aims to give finance and engineering a shared view of how AI budgets are being used. That can make it easier to investigate changes early and assign accountability before bills arrive.

The service is now available to all Ramp customers.

"We're not even live on Ramp yet and we're already using this feature. Everyone wanted AI access, and at the same time, tokens became more expensive, so visibility into AI costs quickly became business critical. The timing couldn't have been better," said Greg K., Director of Corporate IT & Enterprise Security at Front.