Rippleing launches a tool for human resources that links AI spending to employee productivity
Listen to this article
Read by Anchor
When the CFO at Rippleing presented a shocking figure at an early-year executive meeting, the company realized it was on track to spend 40% of the research and development team's salary budget on AI tokens alone. The spending was not just a large number; it was growing by 80% each month, and at that rate the token cost will consume 90% of the compensation budget within a year.
The response was not to halt usage, but to build an internal gateway that routes each request to the most cost-effective and performant model, with dashboards that link each engineer’s and team’s spending to their actual productivity (lines of code, pull requests, registered customers). The result: token cost fell from 40% to 15% of the salary budget, while usage volume remained at 600 billion tokens per month, but at only 37% of the cost it was at during the peak warning.
Tool:"AI Spend Console", included in Rippleing’s human-resources subscription and also available as a standalone product that integrates with other HRIS systems.
Mechanism:The company negotiated a spending cap with each provider (Cohere, OpenAI, Anthropic), then built its own routing gateway that selects the optimal model for each task, discovering, for example, that the Chinese "JLM 5.2" delivers performance comparable to leading models at roughly one-third the cost.
What this means for the region:Companies in the Gulf, Egypt and Morocco that adopt AI coding tools (GitHub Copilot, Cohere, Claude Code) face the same curve: rising spending without governance. A product like this turns "token-maxing" from a financial chaos into an administrative decision measured by work output rather than by the number of calls.
Conclusion:Technology alone is not enough; Rippleing appointed "AI leaders" from the effective engineers to train their peers, and found that engineers are the primary users so far, while sales and customer-support teams are still at the early stage. If they cannot link token consumption to productivity in administrative and commercial roles, then "all bets fall" on opening access to everyone, as product manager Matt McKinney said.