After Rippling blew millions on AI in months, it built an employee ROI tool

In the early months of 2024, the corporate world was swept up in a "tokenmaxxing" frenzy, with companies racing to integrate generative AI into every facet of their operations. Rippling, the HR and payroll software giant, was no exception. However, what began as an ambitious push for technological advancement quickly devolved into a fiscal nightmare. This week, the company unveiled its solution to the chaos: the AI Spend Console, a comprehensive product designed to track, manage, and optimize AI expenditures across teams and individual employees.

The Wake-Up Call: A Multi-Million Dollar Burn

The impetus for the new tool was a sobering realization during an executive meeting this past March. CFO Adam Swiecicki presented data that left the leadership team in disbelief: Rippling was on a trajectory to consume 40% of its entire R&D headcount budget on AI tokens. To put that in perspective, the company was spending nearly as much on digital tokens as it was on the base compensation for its high-paid engineering staff.

With monthly spending growth hitting 80%, projections indicated that within a year, AI costs would balloon to 90% of the R&D payroll. Chief Product Officer Matt MacInnis described the atmosphere as one of total incredulity. The company’s response was immediate and drastic, launching an urgent internal audit to determine exactly what value—if any—was being generated by this massive capital outflow.

"We were incredulous," said CPO Matt MacInnis. "The truth is that the inference providers, like Anthropic and OpenAI, have absolutely no incentives to help you control your spend. They have every incentive for it to be a runaway expense."

Identifying the "AI Slop"

The internal audit revealed a stark reality: the spending was highly concentrated. Rippling discovered that roughly 10% to 15% of its workforce was responsible for 60% of the total AI bill. In one extreme instance, a single engineer was burning through $50,000 in tokens every month.

The AI Spend Console was built to address these inefficiencies by mapping spending directly to output. The tool doesn't just track costs; it correlates them with productivity metrics. It identifies which engineers are racking up high bills while simultaneously requiring their peers to constantly redo their work during code reviews—a phenomenon the company refers to as producing "AI slop."

Key Features of the AI Spend Console:

• Granular Tracking: Monitors AI spend by individual, team, and specific job role.
• Productivity Mapping: Links token consumption to tangible outputs, such as lines of code or completed pull requests.
• AI Gateway: An integrated routing system that directs prompts to the most cost-effective model capable of handling the specific task.
• Performance Dashboards: Visualizes the relationship between spending and actual work output.

The Shift to Strategic Routing

Rippling’s strategy was never to ban AI, but to govern it. The company began by negotiating strict spending caps with providers like OpenAI, Anthropic, and Cursor. They quickly realized that employees were defaulting to the most expensive frontier models for every task, regardless of complexity.

The company has since pivoted to a multi-model approach. By utilizing an internal AI gateway, Rippling now routes prompts to the most efficient model for the job. For instance, while Grok (via Cursor) has proven to be a top-tier performer, the company found that models like Z.ai’s GLM 5.2 offer near-identical performance for coding tasks at a fraction of the cost.

The results of this shift have been dramatic. In July, Rippling’s internal AI usage returned to the same volume as its peak month—600 billion tokens—but the cost was only 37% of what it had been in April.

Beyond Engineering: The Future of AI Productivity

While software engineers have been the primary users of AI thus far, Rippling is actively expanding the tool's reach to other departments, including customer onboarding and G&A functions. The company is appointing "AI captains"—employees who have demonstrated high proficiency—to mentor their peers and ensure that AI usage translates into measurable business value.

"We have to be able to link token consumption in G&A functions and in customer-facing functions back to productivity," MacInnis emphasized. "If we can’t do that, all bets are off on any of this stuff being available to the broader employee base."

This philosophy suggests a significant shift in corporate culture. AI access, once treated as a ubiquitous utility like email or Slack, is now being treated as a resource that must be earned through demonstrated ROI.

For current Rippling HR subscribers, the AI Spend Console is included as part of the platform, though usage-based costs still apply. For companies not using Rippling’s full suite, the console is available as a standalone product that can integrate with other HR systems of record. As enterprises continue to grapple with the ballooning costs of the AI revolution, Rippling’s move to treat AI as a managed asset rather than a blank check may well become the industry standard.