Rippling has launched a tool that tracks corporate AI spending after discovering its own token costs were on course to equal 40% of its research and development payroll. AI Spend Console connects spending with employee, team and role data to help companies assess whether higher AI usage is producing measurable work.
The problem became clear to Rippling executives in March, when CFO Adam Swiecicki reported that AI spending was increasing by about 80% each month. If that continued, annual token costs could have reached roughly 90% of what the company spent compensating its R&D employees.
Rippling found that about 10% to 15% of employees generated roughly 60% of its total AI spending. One engineer alone was spending $50,000 per month, according to the company’s official AI Spend Console announcement.
Rippling Links AI Spending With Employee Output
AI Spend Console breaks down costs across employees, teams and roles, then compares usage with work output. For engineers, for example, the system can combine AI spending and prompts with measures such as lines of code and pull requests, while also identifying cases where peers frequently request that an employee redo work during code reviews.
Rippling initially responded to its rising costs by negotiating maximum spending limits with AI providers including OpenAI, Anthropic and Cursor. The company also found that employees frequently chose the newest and most expensive AI models even when less costly models could handle the same work.
The company subsequently built an AI gateway that routes requests toward models based on the task and cost. Companies can use AI Spend Console alongside another gateway, although Rippling says its own gateway is required for features that directly govern spending.
Rippling CEO Parker Conrad also said the company’s internal testing found that GLM 5.2 delivered nearly identical performance to leading frontier models at 85% lower cost. His comments formed part of Rippling’s case for using multiple models at different price levels rather than relying on one provider for every task.
Token Usage Stayed High While Costs Fell
Rippling said the changes reduced AI spending from the equivalent of 40% of its R&D headcount budget to about 15%. The reduction did not come from substantially cutting AI usage.
Employees consumed 605 billion tokens during the month when management first raised concerns. By July, usage had returned to about 600 billion tokens, but the cost was only 37% of what the company had paid for its April token consumption, according to Chief Product Officer Matt MacInnis.
Rippling has also appointed employees who use AI effectively as “AI captains” to help colleagues. The company is testing ways to measure AI productivity outside engineering, including customer onboarding work where output can be linked to the number of customers processed.
MacInnis said access to AI across general and administrative or customer-facing teams would depend on whether Rippling could connect token consumption with productivity. AI Spend Console is included for Rippling HR customers with additional usage-based AI costs, and Rippling says it can also be purchased separately and connected to another HR system.
Featured image credits: Magnific.com
For more stories like it, click the +Follow button at the top of this page to follow us.