Rippling Spent Millions on AI. Now It Has Built a Tool to Measure Employee ROI

Companies are rushing to adopt AI, but one major question is becoming increasingly difficult to ignore: Is the productivity gained from AI actually worth the cost?

Rippling, the workforce management and business operations company, encountered that question firsthand after its internal AI usage grew rapidly.

According to reporting by TechCrunch, Rippling was on track to spend an amount equivalent to roughly 40% of its R&D employee compensation on AI model tokens. Its AI spending was also growing by about 80% month over month. If that pace had continued, the company's annual AI token bill could have approached 90% of the compensation cost of its R&D workforce.

The experience prompted Rippling to build a tool designed to answer another question: What is the actual return employees are getting from AI?

AI Costs Can Grow Faster Than Expected

Generative AI can look inexpensive when individual interactions cost only fractions of a dollar. The economics change dramatically when thousands of employees and software agents use AI continuously.

Developers can consume large numbers of tokens while coding, testing and debugging. Customer-support teams can use AI throughout the day. Internal tools can make automated model calls in the background.

As AI becomes embedded into everyday workflows, companies can move from experimenting with AI to operating significant AI infrastructure almost overnight. Rippling's experience illustrates this transition.

The company discovered that simply encouraging employees to use AI wasn't enough. It needed to understand whether that usage was producing enough value to justify the rapidly increasing costs.

Measuring AI Productivity Is Hard

Traditional employee productivity metrics are relatively straightforward. A company can measure sales generated, support tickets resolved, projects completed or hours worked.

AI complicates the calculation. An employee might use an AI coding assistant for several hours but spend some of that time reviewing or correcting the generated code. Another employee might use AI for 15 minutes and save several hours of manual work.

Simply counting AI usage therefore doesn't necessarily tell a company whether the technology is producing meaningful economic value. The challenge becomes even greater when AI agents perform tasks automatically in the background.

Rippling's Response: Measure AI's Return

Rather than simply attempting to reduce its AI bill, Rippling developed an internal tool to examine the relationship between AI usage and employee productivity.

The objective is to move the conversation from "How much AI are our employees using?" to "What value are we getting from that usage?"

That distinction could become increasingly important as companies deploy AI across entire departments. Rippling already operates a platform spanning HR, IT and finance, and the company is developing AI infrastructure intended to automate work across those areas.

The New AI Question for Businesses

For years, businesses asked whether they should adopt AI. That question is becoming less relevant. The more important questions are now:

• Which employees should use AI?
• Which tasks should be automated?
• Which AI models should be used?
• How much does each workflow cost?
• How much time does AI save?
• Does AI improve output quality?
• Does the productivity gain justify the infrastructure cost?

These questions turn AI adoption into a business-performance issue, rather than simply an IT initiative.

Not All AI Usage Is Equal

A company could spend $100,000 on AI and receive very little business value. Another company could spend $100,000 and generate several times that amount in additional productivity. The difference may come down to how AI is deployed.

Using expensive models for simple tasks can unnecessarily increase costs. Similarly, allowing employees to use AI without clear workflows can result in duplicated work, excessive model usage and inconsistent results.

Businesses can potentially improve their AI economics by:

• Matching different tasks with appropriate AI models.
• Monitoring token and API consumption.
• Automating repetitive workflows rather than individual tasks.
• Measuring time saved.
• Tracking output quality.
• Establishing usage policies.
• Connecting AI directly to business systems.
• Measuring the financial impact of AI-enabled processes.

Rippling's developer platform, for example, provides APIs, workflows, custom applications and integrations intended to connect business data and automation.

Why This Matters for Smaller Businesses

The lesson isn't limited to large technology companies. Small businesses are increasingly using AI tools for marketing, customer service, sales, administration, research, coding and content production.

But smaller companies may have an even greater need to understand the economics because their technology budgets are limited. A business might subscribe to several AI platforms, pay for automation software, use API-based AI services and employ AI-powered applications without ever calculating the total monthly cost.

That can create a new form of AI tool sprawl. Instead of asking whether an individual tool is affordable, businesses should evaluate the entire AI stack.

What Kenyan Businesses Should Watch

For Kenyan SMEs, the same principle applies as AI adoption accelerates. A business could use AI to automatically respond to WhatsApp enquiries, qualify leads, generate quotations, update a CRM, prepare reports and follow up with customers.

The important question isn't simply whether those tasks can be automated. It is: How much does the automation cost compared with the human time and business value it replaces or creates?

For example, if an automated customer-support workflow costs KSh 20,000 per month but saves a business 100 hours of administrative work and increases conversions, it may be worthwhile. If another AI workflow costs KSh 20,000 but produces little measurable benefit, the business should reconsider it.

That is the type of calculation businesses will increasingly need to make.

AI ROI Could Become a New Business Metric

As AI becomes embedded throughout companies, traditional software metrics may not be enough. Businesses could increasingly track metrics such as:

• AI cost per task — how much it costs to complete a particular AI-powered task.
• Time saved — how many employee hours are eliminated or reduced.
• Cost per outcome — what it costs to generate a qualified lead, resolve a support issue or complete a report.
• AI-assisted revenue — how much additional revenue can be associated with AI-enabled processes.
• Automation ROI — how much financial value is generated compared with the cost of operating the automation.

These measurements can help companies distinguish between AI adoption and AI value.

The Bigger Lesson

Rippling's experience highlights a problem that many companies may encounter as AI becomes cheaper and easier to deploy. The danger isn't necessarily that AI will be too expensive. It may be that AI becomes so easy to use that companies lose track of how much they are spending on it.

Employees can experiment with multiple models. Developers can build AI-powered features. Automated agents can run continuously. Background processes can make thousands of API calls. Without monitoring, costs can grow faster than expected.

Our Take

The next stage of enterprise AI won't simply be about deploying more models. It will be about deploying the right AI, on the right tasks, at the right cost — and proving that it creates measurable value.

Rippling's experience is therefore an important warning for businesses rushing into AI adoption. Don't measure success by the number of employees using AI. Measure the business outcomes produced by AI.

For companies building automation systems, this means ROI should be designed into the workflow from the beginning. Before automating a process, businesses should know how much the existing process costs, what the automation will cost and what measurable improvement is expected.

The companies that master that calculation may gain a significant advantage over businesses that simply adopt AI because everyone else is doing it.

Source: TechCrunch reporting on Rippling's AI spending and employee ROI tool, supplemented by information from Rippling's developer and company resources.