After spending millions on AI, Rippling launches employee ROI tool

Rippling, a leading HR software provider, has launched AI Spend Console, a tool designed to curb tokenmaxxing by helping organizations monitor and control their AI expenditures. A standout feature allows companies to track spending by individual employees, teams, and roles, determining whether higher costs correlate with genuine productivity or merely increased AI-generated noise.

According to the company’s blog post, the tool identifies “engineers with high AI spend whose peers frequently request code rewrites,” highlighting inefficiencies in development workflows.

The solution emerged after Rippling embraced tokenmaxxing early this year, mirroring industry trends, only to find employees rapidly depleting budgets. Chief Product Officer Matt MacInnis recalls a March executive meeting where CFO Adam Swiecicki revealed alarming figures that shocked the leadership team.

Rippling was projected to spend 40% of its R&D headcount budget on AI tokens, equating to spending as much on tokens as 40% of the compensation paid to employees in that department—amounting to millions of dollars. (The R&D division typically houses engineering teams at most tech firms.)

With spending growing at 80% month-over-month, continuing this trend would result in next year’s AI token costs reaching 90% of the budget allocated for high-paid R&D staff.

“We were incredulous,” MacInnis told TechCrunch.

Management immediately initiated an “urgent” project to analyze spending and assess the value derived from these expenses, he explained. The product’s launch advertisement even depicts Swiecicki sitting on a stool while employees toss bundles of cash into a paper shredder.

Upon analysis, Rippling uncovered insights such as “roughly 10–15% of our employees driving about 60% of total AI spend. One engineer was spending $50,000 a month,” according to its blog post.

Rippling did not intend to halt AI usage but rather to significantly rein it in. The company began by negotiating spending caps with its primary tools: Cursor, OpenAI, and Anthropic. This revealed a clear issue: employees defaulted to using the latest, most expensive frontier models for all tasks.

“The truth is that inference providers like Anthropic and OpenAI have absolutely no incentives to help you control your spend. They benefit from runaway expenses, which is exactly what occurs. They fail to provide robust usage insights and do not collaborate with one another,” MacInnis stated.

This was a common challenge in early 2026. Eight months into the year, enterprises have realized two key points: first, they require multiple models from various AI labs at different price points, including frontier open-weight options, potentially from Chinese providers.

Rippling founder and CEO Parker Conrad noted last month that internal benchmarks revealed SpaceX’s Grok as the overall leader, yet “GLM 5.2 is 85% cheaper but [delivered] nearly identical performance” to frontier models. (SpaceX now owns Cursor, which provides access to Grok and dozens of other models.) Z.ai’s GLM 5.2 has emerged as a preferred Chinese model for coding tasks among tech firms, with Databricks also advocating for its use.

Second, enterprises now recognize the need for an AI gateway that directs prompts to the most cost-effective model for each task. Rippling reached this conclusion and built its own AI gateway, integrated into this product. MacInnis noted that enterprises using other gateways can still utilize AI Spend Console, though those seeking spending governance features must use Rippling’s gateway.

AI Spend Console generates dashboards (formerly known as leaderboards during the tokenmaxxing era) that score metrics such as daily prompts combined with work output (lines of code/pull requests) and expenditure.

With this tool, Rippling reduced its token spend from 40% of its headcount budget to approximately 15%, without curtailing AI usage. The company peaked at 605 billion tokens in the month the CFO issued his warning, MacInnis shared. 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,” he added.

“That’s simply because we are now routing to more effective models,” he said, joking that “we’re not allowing the sales team to perform grammar updates using Fable.”

However, Rippling emphasizes that technology alone is insufficient. The company identified employees using AI effectively and designated them as “AI captains” to assist the rest of the organization.

MacInnis noted that efforts to extend AI usage beyond engineering remain a work in progress, as software engineers have been the primary users thus far. Nevertheless, Rippling is applying this approach to customer onboarding teams to automate mailing data and reconciliation tasks. The dashboard will then measure productivity based on the number of onboarded customers.

“We must be able to link token consumption in G&A and customer-facing functions back to productivity. If we cannot do this, the availability of these tools to the broader employee base is uncertain,” MacInnis stated.

Thus, if Rippling serves as an example, tokenmaxxing may have swung so far in the opposite direction that employee AI access may no longer resemble the ubiquity of Slack or email. If productivity cannot be measured, not all employees may retain access.

Regarding the product, AI Spend Console is included for Rippling’s HR subscribers, though additional usage-based costs apply. It can also be purchased as a standalone product and integrated with other HR systems of record, MacInnis stated.