Skip to content
Fast Company Executive Board

The Fast Company Executive Board is a private, fee-based network of influential leaders, experts, executives, and entrepreneurs who share their insights with our audience.

Here’s why your employees don’t use your AI licenses

There’s a massive divergence between corporate AI and personal AI.

[Photo: Getty Images]

Profile photo of Russ Fradin

CEO · Larridin

There is a secret in the IT budgets of almost every large public company in the world right now.

On paper, the strategy looks sound. The CIO has signed a massive enterprise agreement, likely with Microsoft for Copilot, or perhaps Salesforce or OpenAI Enterprise. The company pays $30 to $80 per seat every month. It has rolled out the official secure tools and ticked the compliance box.

But if you walk the floor of that same company and look at what the teams who perform their best work are actually doing, you’ll probably see a different story.

The high performing teams are alt-tabbing away from their corporate-funded AI app to log into their personal ChatGPT Plus accounts or to use Perplexity on their personal phones. They are paying $20 a month out of their own pockets because the tools they use at home are faster, smarter, and less restricted than the applications offered at work.

There’s a massive divergence between corporate AI (what IT bought) and personal AI (what is actually doing the work). As a result, most leaders are flying blind because they can’t measure the right things.

The visibility gap

In previous technological shifts, companies tracked assets: the number of laptops, server pings, and software licenses. AI is not software, though; AI is a workforce so it’s harder to track.

When employees choose to use a personal account over the corporate one, they aren’t just using an unauthorized app. They are signaling a productivity gap. They are telling their leaders that the company tool is failing to meet their needs.

The problem is that traditional network security tools can’t see this. They see traffic flowing to an OpenAI API but cannot distinguish between a sanctioned enterprise query and a user pasting proprietary code into a personal account. Also, they cannot measure if that interaction was a frustrated user trying five times to get an answer, or a power user automating a four-hour task in three minutes.

The strategy deficit

This lack of visibility is, of course, a security risk. But mostly it’s a strategic failure. Many companies are freezing new AI initiatives because they simply don’t know what is currently running on their networks.

I recently discussed this with Karyn Price, industry director for AI at Frost & Sullivan, a global consulting and market research firm. She explained why this paralysis may become common across major industries.

“Most companies are still in the immature, exploratory phases of AI adoption,” Price said. While many are moving forward with technology implementation and data readiness, leaders often lack a clear, deliberate roadmap for why, where, and how AI will be deployed.

“Leaders are implementing technology, but they lack the measurement frameworks to understand if that technology is actually driving value, or to get their arms around the ‘shadow’ usage already happening in their organization,” Price explained.

Price is right. Organizations cannot build a roadmap if they don’t know where they are starting. In fact, they cannot manage what they cannot measure.

AI needs independent measurement

History tells us that whenever a massive amount of capital moves into a new category, independent measurement becomes a requirement.

When TV advertising exploded, Nielsen was needed, and when the internet became a business, Comscore came into play. The same logic applies to AI.

When leaders rely solely on vendors for analytics, they will get answers that serve the vendor. Any company will say its product is essential and performing great.

However, who tells leaders that the marketing team is actually 30% more productive using a $20 a month subscription than they are with the expensive enterprise suite? Or that developers have abandoned the official coding assistant for a newer, faster tool?

Companies need a measurement that sits at the browser and desktop level—the level where all work happens. They need metrics to understand three elements:

1. Utilization: Not just who has a license, but who is active daily.

    2. Proficiency: Who is a beginner using AI as a glorified spellchecker, and who is an advanced user able to build AI agents?

    3. The Split: What is the ratio of enterprise usage vs. personal usage?

      The goal isn’t to create a surveillance state or to block every unauthorized tool. That is a losing battle. The goal is building execution intelligence.

      If a relevant percentage of the workforce prefers Perplexity over the internal search tool, don’t block it. Ask why. Is it better? Is it cheaper? Should the enterprise buy a license for that instead?

      Companies are moving toward a three-currency enterprise where leaders must allocate resources between U.S. talent, remote talent, and AI agents. To navigate this challenge, they need to stop measuring licenses and start measuring work.

      We are in the middle of the largest capital deployment in corporate history, yet most business leaders are measuring it with tools built for the cloud era. It’s time to accept the reality of the personal AI economy before falling behind.

      Russ Fradin is CEO of Larridin.