An Italian startup called Nebuly S.r.l. says it’s going to help organizations better understand the impact their investments in artificial intelligence have on their bottom lines after raising $21 million in funding today.

The round was led by Planven and saw participation from backers including Exor Ventures, Tenacity Capital, TGVG Capital, Fastweb, Vodafone and CDP Venture Capital’s Digital Transition Fund.

The world’s biggest companies have been spending millions of dollars on artificial intelligence, but very few can actually say how much value those investments have returned. Though some leaders might be able to point to a few vanity metrics, they’ll likely stumble if pressed to come up with concrete numbers.

This is a problem because enterprise leaders are increasingly feeling the pressure to justify their investments in AI. A recent Gartner Inc. survey of business leaders found that 85% of respondents plan to step up their AI spending in the next year, but at the same time, only 22% say they have managed to scale the technology across multiple business units.

Meanwhile, spending is spiraling out of control at many companies. Famously, Uber Technologies Inc. revealed earlier this year that it had eaten up the entire budget it set aside for AI in 2026 in just four months.

Nebuly co-founder and Chief Executive Franceso Signorato told SiliconANGLE that one of the main challenges when it comes to understanding AI’s return on investment is that many companies have deployed a jumbled patchwork of AI models and applications. They’re using tools from OpenAI Group PBC, Anthropic PBC, Microsoft Corp., Google LLC and others, as well as developing in-house tools under their own steam.

That makes it harder to see which projects are actually paying off and generating real business revenue. Although companies can track superficial metrics like token consumption and prompt tallies, genuine productivity gains remain obscured.

“Every enterprise is about to face the same question from its board, if it hasn’t already: What is AI actually worth to us? Nebuly exists to answer that question,” Signorato said. “Until now companies have measured AI by counting users, prompts and responses, which has fueled the rise of tokenmaxxing, where usage itself becomes the goal.”

Though it can be useful to keep track of who is using AI the most, Signarato said it doesn’t reveal if the technology is actually providing a net benefit to the company. “It doesn’t tell you whether an employee got back an hour of work or if a customer got the answer they needed,” he explained.

The lack of real insights can be risky, because AI is already starting to influence how business strategies and how companies spend money in other areas. “Companies are making decisions about hiring and head counts based on assumptions about AI productivity, not evidence,” Signorato said. “Decisions of that weight need real data.”

So it’s just as well that Nebuly claims to have the solution for these organizations. It has created an AI measurement platform that connects directly to the AI agents operating within its customer’s information technology environments, so it can study their behavior through a unified control plane.

Nebuly seeks to evaluate agentic productivity across two dimensions. In the case of internal, employee-facing agents, it tracks operational efficiency metrics, such as the number of hours of work they save human employees and the amount of capacity they free up.

It also takes into account net-new work that human teams couldn’t previously do. In addition, it identifies the kinds of work staff are applying AI to – such as routine, mundane tasks or higher-level strategic work, in order to help companies understand where their agents are most useful.

The platform also tracks external, customer-facing agents. With them, it monitors how they influence revenue, their impact on customer churn and their ability to generate upsell opportunities.

Auditability is a key concern, and so Nebuly ensures flexibility in terms of the financial assumptions it makes. This means financial teams can dig in and inspect the math it has come up with, to verify the performance insights it generates. The platform can be deployed on-premises or in private cloud environments in order to keep sensitive business data secure.

Nebuly’s insights are especially appealing to larger organizations. Almost all of its customers pull in annual revenue of more than $100 million, with about half exceeding $1 billion per year. They include Carl Zeiss AG, Iveco Group N.V. and Oura Health Oy. Most of those customers are based in Europe, but the funds from today’s round should help to change that, with Signorato planning to increase go-to-market hiring in the U.S.

Planven’s Giovanni Canetta Roeder said he’s backing Nebuly because he’s convinced that every large organization is going to need a way to measure AI’s return on investment. “Nebuly is building the way to do it,” he said. “The team turns what AI agents actually do into evidence of business value. That capability and the traction Nebuly has built among major enterprises made this a compelling investment for us.”

Image: Nebuly