Rippling caught one engineer burning $50K/month on AI — now it's selling the fix

Rippling just productized its own panic attack. The HR software company launched AI Spend Console this week, a tool that maps exactly how much cash individual employees, teams, and roles are torching on AI tokens — and whether any of that spending produces actual work. The product was born from a brutal internal discovery: Rippling was on track to burn 40% of its R&D headcount budget on AI tokens this year. Millions of dollars. One engineer alone was spending $50,000 a month.

The wake-up call came in March, when CFO Adam Swiecicki dropped a number that Chief Product Officer Matt MacInnis described as making the executive team “incredulous.” AI spending was growing 80% month-over-month. Left unchecked, the company would spend nearly as much on tokens — 90% — as it did on all compensation for its R&D unit within a year. The launch ad for the new product leans into the absurdity, featuring Swiecicki on a stool while employees feed wads of cash into a paper shredder.

Rippling’s subsequent audit revealed that 10-15% of employees drove roughly 60% of total AI spend. The problem wasn’t just volume; it was laziness. Employees defaulted to the most expensive frontier models for every task, from complex code generation to grammar fixes. MacInnis pointed a finger squarely at inference providers: “Anthropic and OpenAI have absolutely no incentives to help you control your spend.” So Rippling built its own AI gateway that routes prompts to the most cost-effective model for the job. The company found that Z.ai’s GLM 5.2, a Chinese model, delivered nearly identical performance to frontier models at 85% less cost. SpaceX’s Grok also emerged as an all-around leader in internal benchmarks.

With the routing in place, Rippling slashed its token spend from 40% of its headcount budget to about 15% — without actually reducing AI usage. In July, internal token consumption hit 600 billion again, the same peak recorded during the CFO’s warning month, but the cost was 37% lower. The tool’s dashboards score attributes like prompts per day alongside work output, and the company isn’t shy about the social pressure this creates. The system can flag “which engineers have high AI spend whose peers frequently ask them to redo work in code reviews.” Rippling also appointed effective AI users as “AI captains” to coach the rest of the company, an acknowledgment that technology alone won’t fix a culture problem.