Case studies · Enterprise

A global enterprise puts a cost and an owner on each AI call

Tokens and cost recorded on each call through Raidu, by team, model and decision, with a budget alert before spend runs over.

TowerAI spendCost and owner per call
Calls through Raidu, last minuteTotal $0.00
  • Claimsclaims-assistant · Claude Sonnet18,240 tokens$0.42
  • MarketingAI Console · GPT-4.16,410 tokens$0.09
  • EngineeringCursor · Claude Sonnet41,900 tokens$0.96
  • FinanceAI Console · Gemini3,120 tokens$0.03
Owner and cost on each record
Illustrative product view. Sample data, not client data.

The problem

AI cost arrives as a handful of provider invoices. Nobody can say which team, which model or which decision a dollar went to, and finance finds out at month end.

  • Per calltokens and cost on the record
  • 80%budget alert to the team lead
  • Alertsnot hard stops: work carries on

Before Raidu

  • Provider invoices with no owner
  • No budget per team
  • Spend discovered at month end

With Raidu

  • Each team sees its own AI spend
  • An alert at 80% of budget
  • Cost by team, model and decision on the record

The challenge

The company has thousands of people using several model providers through chat, internal apps and agents. Each provider bills separately. Finance can see the total, but not who spent it or on what, and budget conversations happen after the money is gone.

How they set it up

Every approved app and agent sends its model calls through Raidu in The company’s own cloud. Each call is checked against policy, and the record keeps the team, the person or agent, the model, the input, output and cached tokens, and the cost. Team leads get an alert when their team passes 80% of its monthly budget. Nothing is blocked for cost; work carries on and the conversation happens early.

What finance gets

A monthly view of AI spend by team, model and decision, taken from the same signed record security uses. When a cost spikes, they can see which app and which call caused it.

Rollout

  1. Step 1Raidu installed in their AWS account; provider keys moved behind it
  2. Step 2Budgets set per team; alerts routed to team leads
  3. Step 4Finance reviews spend by team and model from the record

What they deployed

  • AI Spend: a budget for each team, group or person, with an alert at 80%
  • AI Console: one approved place for the models people use
  • Evidence and Audit: input, output and cached tokens and cost on each record

Talk to us about your data and AI.

Tell us what you need to prove and to whom. We will show you the record it produces.

Talk to us