Enterprise AI spending has reached an average of $186 million annually, yet most organizations struggle to measure whether their AI investments are actually delivering business value. As companies pour billions into artificial intelligence (AI) tools and agents, a critical gap has emerged: they can track how much they're spending, but not what they're getting in return.

Why Can't Companies Measure AI ROI?

The disconnect is stark. A recent KPMG report found that only 8% of enterprises have achieved meaningful business returns with AI, while Deloitte reported that 74% of organizations want AI to grow revenue, but only 20% have seen it happen. As budgets tighten and executives demand accountability, what was once a headline about "AI sticker shock" has moved into serious boardroom conversations.

The problem, according to Botanu, a startup that emerged from stealth on June 11, 2026, isn't that AI doesn't work. It's that companies can't locate where their AI agents are actually working or measure the outcomes they produce. "This isn't a bubble. It's a measurement problem," said Alina Vrsaljko, Botanu's co-founder and CEO.

"Companies aren't failing because AI doesn't work. They're failing because they can't locate where their agents are working. And this isn't a tech problem anymore,72% of CEOs now own the AI decision, and they have no way to prove it's paying off," said Vrsaljko.

Alina Vrsaljko, CEO and Co-founder, Botanu

The core issue is structural. A single AI agent's costs are scattered across multiple systems, each metered differently and owned by different teams. Meanwhile, the value that agent creates is equally fragmented, spread across customer relationship management (CRM) systems, support tickets, and revenue records. No single dashboard connects the dots between what an AI agent costs and what it actually delivered.

How Are Companies Currently Measuring AI Success?

Most organizations are measuring the wrong metrics. They track activity,tokens consumed, tool calls made, login frequency,rather than outcomes. This creates what experts call "token-maxxing," where companies celebrate high usage numbers without understanding whether those activities translate into business value.

Ray Rike, CEO and co-founder of Benchmarkit, explained the fundamental disconnect: "Token spend is up 13x since January 2025, yet only 27% of executives say AI has met their ROI expectations. The problem is that adoption is not the same as value creation. Companies are celebrating adoption metrics, tokens consumed and agents deployed, while the value created stays invisible".

Ray Rike, CEO and co-founder of Benchmarkit

"Enterprises are running out of budget before they run out of enthusiasm. The discipline that's missing is simple to say and hard to do: measure outcomes, not activity, and connect costs to returns," said Ray Rike.

Ray Rike, CEO and Co-founder, Benchmarkit

The shift in how AI is priced has made this problem worse. In the cloud computing era, costs increased predictably with usage, and every bill could be traced back to the specific workload or team that drove it. AI breaks that model. The same task can produce vastly different costs from one run to the next, with little predictability upfront. Usage-based pricing models push cost volatility directly onto the buyer's invoice, making it nearly impossible to know which agents were worth the investment.

What Does Outcome-Based Measurement Look Like?

The emerging solution is to treat AI agents like employees rather than software licenses. "An AI agent is a new kind of workforce, and it works at 100 times the frequency of a person," Vrsaljko explained. "You should performance-manage it, not just cost-manage it. The question isn't 'Why is this so expensive?' It's 'Is this agent doing the job, and is the job worth the salary?'".

This reframing requires connecting two halves of the equation that most tools ignore: cost and value. Deborah Jacob, Botanu's co-founder and chief technology officer, emphasized the importance of measuring actual business results rather than activity proxies.

"Activity is not outcome. A thousand tokens and ten tool calls tell you an agent was busy,not whether it closed the deal. We measure the result the business actually recorded, and weigh it against what it cost to get there,the one number a CFO can act on," said Jacob.

Deborah Jacob, Chief Technology Officer and Co-founder, Botanu

Steps to Connect AI Spending to Business Outcomes

• Map Agent Costs Across Systems: Track where an AI agent's expenses appear across different platforms, vendors, and infrastructure layers, not just token consumption, to understand the true cost of deployment.
• Link Outcomes to Business Systems: Connect AI agent activity to actual business results recorded in systems like CRMs, support platforms, and revenue tracking tools to measure real impact.
• Compare to Human Labor Costs: Benchmark AI agent performance against what the same job would cost if performed by a human employee to determine whether the investment makes financial sense.
• Measure Results, Not Activity: Focus on outcomes,qualified leads generated, tickets resolved, revenue influenced,rather than activity metrics like tokens consumed or tool calls made.
• Establish CFO-Level Governance: Create accountability frameworks that allow finance leaders to see which AI investments deliver 3x returns and which ones should be cut or scaled.

The stakes are rising. As enterprises continue to increase AI spending at roughly one-third growth year over year, budget planning cycles will demand accountability. For some companies, AI spending already exceeds 1% of total revenue. Leaders who can't answer what they got for their investment will face difficult questions from boards and CFOs.

What's Driving the Shift Toward Outcome Measurement?

The enterprise AI story has fundamentally shifted. For two years, the focus was on adoption: how many pilots could a company launch, and how quickly could it roll out tools? In 2026, the boardroom question has flipped entirely. It's no longer about how much a company spends on AI, but what it got back.

This shift is being driven by CFO scrutiny and the realization that most organizations lack the governance frameworks to manage AI spending like any other major investment. M.G. Thibault, who leads the Coterie CFO community and serves as CFO-in-residence at Scale Venture Partners, noted the gap: "We're spending confidently on AI. What we're missing is a way to measure it that every CFO would recognize; a real KPI, not usage stats. That's the open space right now".

Meanwhile, other organizations are taking a different approach to workforce transformation. LTM, a global technology services company, launched AI 1000 on June 12, 2026, a strategic initiative to develop over 1,000 AI-certified engineers, including Forward Deployed Engineers (FDEs) who combine technical AI knowledge with business understanding to drive measurable ROI. The program uses a four-stage model: Identify high-potential engineers, Enable them through curated learning, Deploy them into AI programs, and Govern their performance through continuous feedback loops.

"The role of the technology engineer is evolving rapidly. AI 1000 is built with the purpose of enhancing workforce productivity in creating tangible business outcomes. Success will be measured not by the number of employees trained, but by the outcomes those engineers deliver," said Venu Lambu, CEO and Managing Director of LTM.

Venu Lambu, CEO and Managing Director, LTM

LTM's approach reflects a broader recognition that AI transformation requires more than tools and training. It requires people who understand both the technical capabilities of AI systems and the business context in which they operate. The company has already trained over 24,000 AI-trained associates and achieved nearly 84% learning penetration across its workforce, with more than 15,000 external AI certifications earned.

The divide in enterprise AI is no longer between companies that spend on AI and those that don't. It's between those that can measure their AI investments and optimize them, and those that can't. Organizations that start with the business outcome in mind,defining what job an AI agent should do and how success will be measured,report getting about three dollars of value for every dollar invested. Right now, only about 6% of companies lead with that discipline. As budgets tighten and accountability increases, that gap will likely widen significantly.