Rippling burned millions in tokens without noticing — and turned the mistake into a product: meet the AI Spend Console

Launched on August 6, 2026, the AI Spend Console connects token spend with employee data, productivity, and business results — because the question is not how much you spend on AI, but what that spend is producing

PrezenceAI Editorial · Published on 08/08/2026

🤖

This article was produced by PrezenceAI with the assistance of artificial intelligence. The main information was extracted from sources cited at the end of this text. The analysis, synthesis, and writing are the sole editorial responsibility of PrezenceAI.

🎧 Listen to this article

0:00 / --:--

The story begins with an internal scare. Rippling — a workforce management platform — discovered in early 2026 that its AI token spend was growing 80% per month. Upon investigating, it found individual engineers consuming the equivalent of 40% of the R&D headcount budget. "We were shocked," said Matt MacInnis, a company executive. In one month, the cost had reached the equivalent of $50,000 per month on a single engineer.

The cause: employees using by default the latest — and most expensive — frontier models for all tasks, regardless of complexity. Rippling had encouraged limitless experimentation. The result was an unexpected bill that forced an urgent review.

What Rippling discovered

The company negotiated spend caps with each tool — Cursor, OpenAI, and Anthropic. Then it discovered another problem: the providers themselves have no incentive to help control costs. "Inference providers have absolutely no incentive to help you control your spend. They have every incentive for it to be an uncontrolled expense," said MacInnis.

The internal solution: route requests to more efficient models. GLM 5.2 — an open-weight model — was found to be 85% cheaper than frontier models with almost identical performance for routine tasks. In July 2026, internal consumption reached 600 billion tokens — similar to the April peak — but the cost was only 37% of the April value. Spend dropped from 40% to 15% of the R&D budget.

What the AI Spend Console is

Launched on August 6, 2026, the AI Spend Console connects token spend with Rippling's Employee Graph — a record of employees, departments, roles, and reporting lines. This allows seeing not only how much each person spends, but if that spend produces a result.

Concrete example: the system can identify engineers with high AI spend and show if their peers frequently ask them to redo work during code reviews. Rippling's CFO Adam Swiecicki was direct: "The question isn't how much you're spending on AI. It's what your AI spend is producing. Until you can answer that, you're just managing costs — not results."

Main capabilities: spend breakdown by employee, team, and role; gateway that routes requests to the most efficient models; connection with GitHub and Salesforce data to measure real productivity; enforcement of spend policies by department.

What this signals

Rippling turned an expensive lesson into a product — and the timing is perfect. The market that in 2025 accumulated AI tools will spend 2026 discovering which ones actually work. The AI Spend Console arrives when CFOs and CTOs are asking exactly this question.