Updated
Updated · TechCrunch · Aug 7
Rippling Launches AI Spend Console After Cutting Token Costs From 40% to 15%
Updated
Updated · TechCrunch · Aug 7

Rippling Launches AI Spend Console After Cutting Token Costs From 40% to 15%

2 articles · Updated · TechCrunch · Aug 7

Summary

  • Rippling this week rolled out AI Spend Console after finding its AI-token bill was on pace to consume 40% of its R&D headcount budget, with spending rising 80% month over month.
  • Its internal review found 10%–15% of employees drove about 60% of total AI spend, including one engineer using $50,000 a month, prompting caps on vendors and closer tracking of output versus cost.
  • The product maps spending by employee, team and role, flags costly low-value usage in areas like code reviews, and ties governance features to Rippling’s own AI gateway that routes prompts to cheaper models.
  • With that routing in place, Rippling said token spend fell to about 15% of headcount budget; July usage still reached 600 billion tokens, but cost was 37% of April’s.
  • Rippling says broader employee AI access may depend on proving productivity outside engineering, as it tests use cases like customer onboarding and sells the tool both to HR subscribers and as a stand-alone product.

Insights

If one engineer burns $50,000 a month on AI, are companies secretly funding bot-driven busywork instead of real breakthroughs?
Could tracking every AI token an employee uses actually destroy the very innovation and experimentation it aims to measure?
Will routing AI tasks to cheaper models quietly degrade your company's code quality before management even notices the difference?