Rippling built employee ROI tool after AI spending spree

HR software firm Rippling has introduced AI Spend Console, a tool designed to help organizations monitor and control their artificial intelligence expenditures. The product stands out for its ability to break down spending by individual employees, teams, and roles, while also assessing whether that investment translates into real productivity gains or simply generates more low-quality AI output.

The company positions the solution as a way to identify which engineers have high AI costs yet whose work frequently gets sent back for revisions during code reviews, according to the company’s announcement.

The development of this product traces back to Rippling’s own aggressive adoption of AI tools at the beginning of the year. Chief Product Officer Matt MacInnis remembers the March executive meeting when CFO Adam Swiecicki revealed a figure that caught everyone off guard. Rippling was on pace to consume 40% of its R&D headcount budget on AI tokens, meaning the company was spending nearly half of what it paid its entire research and development workforce on token usage alone. That translated into millions of dollars.

The situation was worsening rapidly. Monthly spending was climbing by 80%, and projections showed that if left unchecked, the following year would see AI token costs reach 90% of the R&D unit’s total compensation expenses.

“We were incredulous,” MacInnis told TechCrunch.

Leadership launched an urgent investigation to understand where the money was going and what value it was generating. The promotional material for the new product even features Swiecicki sitting on a stool while employees feed cash into a paper shredder, a visual metaphor for the waste they uncovered.

The internal analysis revealed striking patterns. Roughly 10 to 15% of employees accounted for about 60% of all AI spending. One engineer alone was burning through $50,000 per month.

Rippling’s goal was not to eliminate AI usage but to bring it under control. The first step involved negotiating maximum spending caps with each AI provider the company used, including Cursor, OpenAI, and Anthropic. This quickly exposed a fundamental problem: employees were defaulting to the newest, most expensive frontier models for every task, regardless of complexity.

“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, and that’s exactly what they do. They don’t provide you with great usage insight, and they don’t collaborate with one another,” MacInnis said.

This scenario was common in early 2026. By now, eight months into the year, enterprises have learned two key lessons. First, they need access to multiple models from different AI labs at various price points, including open weight options that may come from Chinese providers.

Rippling founder and CEO Parker Conrad noted last month that internal benchmarks for the company’s own use cases showed SpaceX’s Grok performed best overall, but “GLM 5.2 is 85% cheaper but [had] nearly identical performance” compared to frontier models. SpaceX now owns Cursor, which provides access to Grok and numerous other models. Z.ai’s GLM 5.2 has become a preferred Chinese model for coding tasks among many tech companies, with Databricks also promoting it.

Second, enterprises now recognize the need for an AI gateway that routes prompts to the most suitable and cost-effective model for each task. Rippling reached this conclusion as well and built its own gateway as part of the new product. MacInnis explains that companies already using another gateway can still adopt AI Spend Console, though accessing the spending control features would require switching to Rippling’s gateway.

AI Spend Console generates dashboards that score metrics such as daily prompts, work output measured in lines of code and pull requests, and associated costs.

After implementing the tool, Rippling reported that token spending dropped from 40% of its headcount budget to roughly 15%. Notably, this reduction did not come from cutting AI usage. The company hit a peak of 605 billion tokens in the month the CFO issued his warning. In July, internal usage reached 600 billion tokens again, yet “the cost of July’s token spend was 37% of the cost of April’s token spend,” MacInnis shared.

“That’s just because now we’re routing to the more effective models,” he said, adding with humor that “we’re not letting the sales team do grammar updates using Fable.”

Technology alone is not sufficient, Rippling acknowledges. The company identified employees who used AI effectively and designated them as “AI captains” to assist others across the organization.

Still, expanding AI adoption beyond engineering remains a work in progress, MacInnis says, since software engineers have been the primary users so far. Rippling is currently exploring applications for customer onboarding teams, such as automating mailing data and data reconciliation tasks. The dashboard would then measure productivity based on the number of customers onboarded.

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

If Rippling’s experience is any indication, the pendulum may have swung so far that employee access to AI could become more restricted than tools like Slack or email. Without measurable productivity gains, universal access may no longer be guaranteed.

AI Spend Console is included for Rippling’s existing HR subscribers, though additional usage-based AI costs apply. It is also available as a standalone product that can integrate with another HR system of record, according to MacInnis.