When a company burns through millions of dollars on AI tokens in a matter of months, the response tends to be panic, not product launches. For Rippling, an HR and workforce management software provider, that panic turned into a two-pronged project: get its own runaway AI spending under control, and build a tool that other enterprises could use to avoid the same fate. This week, the company unveiled AI Spend Console, an anti-tokenmaxxing product designed to help organizations track, contain, and optimize their spending on large language models. The tool arrives at a moment when enterprises across every sector are grappling with the same question Rippling faced in early 2026: How do you harness the power of AI without letting it devour your budget?
The Tokenmaxxing Wake-Up Call: How Rippling Discovered 40% of Its R&D Budget Was Going to AI Tokens
Rippling went all in on tokenmaxxing at the start of the year, as so many companies did. The logic was simple: give employees unrestricted access to frontier AI models and watch productivity soar. But by March, Chief Financial Officer Adam Swiecicki presented a number to the executive team that left them incredulous. Rippling was on track to burn 40% of its R&D headcount budget on AI tokens. In real terms, that meant the company was spending as much on tokens as 40% of all the compensation it paid to employees in its engineering-heavy R&D unit. Millions of dollars were vanishing each month.
The spending was accelerating at an alarming rate: month-over-month growth of 80%. If that trend continued, the next year would see Rippling spend nearly as much on AI tokens (90%) as it did on its high-paid R&D employees. Chief Product Officer Matt MacInnis recalled the moment vividly. “We were incredulous,” he told TechCrunch. Management immediately launched an “urgent” project to understand where the money was going and what they were getting for it.
What Is AI Spend Console? Rippling’s Answer to Runaway AI Costs
AI Spend Console is a software tool that gives companies granular visibility into how their AI budget is being consumed. It maps spending by individual employees, teams, and roles, and then correlates that spending with productivity metrics. One of its most interesting features is the ability to identify which engineers have high AI spend but whose peers frequently ask them to redo work in code reviews — a practical proxy for the output of “AI slopaa” rather than genuine productivity gains. The tool also provides dashboards (once known as leaderboards during the tokenmaxxing era) that score attributes such as prompts per day combined with work output, like lines of code or pull requests, and spending.
Rippling’s own data revealed a striking pattern: “Roughly 10–15% of our employees were driving about 60% of total AI spend. One engineer was spending $50,000 a month,” the company’s blog post shared. The product is designed to surface such extremes and give managers the ability to rein them in without shutting off AI access entirely.
How Rippling Cut Token Spend from 40% to 15% of Headcount Budget Without Reducing Usage
When Rippling analyzed its AI spend patterns, it discovered several issues. The most glaring was that employees defaulted to using the most recent and most expensive frontier models for all tasks, regardless of whether the task required that level of capability. The company also found that the AI inference providers themselves offered little help. “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.
Rippling’s first move was to negotiate a maximum spending cap with each of the tools its company used: Cursor, OpenAI, and Anthropic. But caps alone were not enough. The company needed a way to route each prompt to the best and most cost-effective model for the specific task. So it built its own AI gateway, which is now a core part of AI Spend Console. The gateway allows Rippling (and its customers) to automatically select a cheaper model — such as Z.ai’s GLM 5.2 — for tasks where performance is nearly identical to the frontier model. CEO Parker Conrad noted last month that when Rippling conducted internal benchmarks, it discovered that SpaceX’s Grok was the all-around leader but that “GLM 5.2 is 85% cheaper but [had] nearly identical performance” to the frontier models.
The results were dramatic. Rippling reported that its token spend dropped from 40% of its headcount budget to about 15%. But it did not curtail AI usage. The company hit a peak of 605 billion tokens the month the CFO issued his warning. As of July, internal usage hit 600 billion tokens again, yet “the cost of July’s token spend was 37% of the cost of April’s token spend,” MacInnis explained. “That’s just because now we’re routing to the more effective models.”
The Role of the AI Gateway: Why Rippling Built Its Own Routing Engine
Central to the spend reduction is the AI gateway that Rippling built. Enterprises that already use another gateway can still adopt the AI Spend Console for visibility, but to get the full spending governance features — including automatic routing and caps — they must use Rippling’s gateway. MacInnis emphasized that the gateway is not just a cost-saver; it is a governance layer that ensures prompts go to the right model based on the task’s complexity and the company’s budget rules. “We’re not letting the sales team do grammar updates using Fable,” he joked, referring to a scenario where an expensive frontier model is used for trivial tasks.
Beyond Engineering: Measuring Productivity Across the Entire Organization
Rippling’s internal efforts to control AI spending are not purely technological. The company found employees who were using AI effectively and designated them as “AI captains” tasked with assisting the rest of the organization. But extending AI beyond engineering is still a work in progress. MacInnis noted that software engineers have been the primary users so far. Rippling is now working on applications for customer onboarding teams, where AI can automate mailing data and data-reconciliation tasks. The dashboard will then measure productivity in terms of onboarding more customers, linking token consumption directly to business outcomes.
MacInnis laid out the challenge clearly: “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.” This statement carries a significant implication for the future of enterprise AI. If other companies follow Rippling’s example, tokenmaxxing may have swung so far in the other direction that employee AI access will no longer be as ubiquitous as Slack or email. If a company cannot measure the productivity gains from AI, it may restrict access to only those who can demonstrate measurable returns.
What AI Spend Console Means for the Enterprise AI Market
Rippling’s product launch comes at a time when enterprises have learned several hard lessons about AI spending. First, they know they need access to multiple models from multiple AI labs at various price points, including frontier open-weight options — possibly of Chinese origin, such as Z.ai’s GLM 5.2, which has become a favorite for coding tasks. Second, they now recognize that an AI gateway is essential for routing prompts efficiently. The market for such gateways is rapidly maturing, with Databricks also championing GLM 5.2 and SpaceX’s Cursor now offering access to Grok and dozens of other models.
Rippling’s AI Spend Console is included for its HR subscribers, though there are additional AI usage-based costs. It can also be purchased as a stand-alone product and integrated with another HR system of record. This flexibility signals that Rippling is positioning the tool not just as an add-on but as a potential industry-wide solution for the spending crisis that nearly every large organization is facing.
As the enterprise AI landscape evolves, the lesson from Rippling is clear: the era of unrestricted token spending is over. Without robust measurement and routing, AI adoption risks becoming a financial liability rather than a productivity accelerator. Rippling’s experience — from the shock of a 40% burn rate to the engineering discipline that brought it down to 15% — offers a blueprint for companies that want to keep their AI ambitions alive while keeping their budgets intact.
