Enterprise AI spending hit $186 million annually per organization in 2026, but most CFOs can't tell you what they got back. The measurement gap is turning boardroom enthusiasm into budget panic. Botanu, a New York- and San Francisco-based startup, emerged from stealth June 11 saying the fix isn't spending less—it's seeing where AI agents actually create value.
The company calls it "outcome-maxxing": measuring what agents deliver, not just what they consume.
The Numbers Behind the Crisis
Only 8% of enterprises have achieved meaningful business returns with AI, according to KPMG. Deloitte found that 74% of organizations want AI to grow revenue, but only 20% have seen it happen. Token spend is up 13x since January 2025, yet only 27% of executives say AI has met their ROI expectations.
The gap isn't adoption. It's accountability. Companies are celebrating tokens consumed and agents deployed while the value created stays invisible.
"Enterprises are running out of budget before they run out of enthusiasm," said Ray Rike, CEO of Benchmarkit and host of the "AI to ROI" podcast. "The discipline that's missing is simple to say and hard to do: measure outcomes, not activity, and connect costs to returns. Until then, AI is a compensation-scale expense that demands CFO-level governance most companies don't yet have."
72% of CEOs now own the AI decision, according to BCG. They have no way to prove it's paying off. The bill arriving isn't the problem—not knowing whether you got $3 of value for every $1 you spent is the problem.
Why the cloud Playbook Breaks
In the cloud era, cost increased with usage and every bill could be mapped back to the workload, team, or product that drove it. AI breaks that model. The same task can produce very different costs from one run to the next, with little predictability upfront. Pricing is shifting from flat per-seat subscriptions to usage-based models. That pushes volatility onto the buyer's invoice.
By the time the bill comes in, no one can tell which agents were actually worth it.
"A single agent's cost is scattered across systems, each metered differently, each owned by a different team. No one can see what one agent actually costs," said Deborah Jacob, Botanu's co-founder and CTO. "But cost is only half the problem. The value an agent creates is just as scattered as its spend."
For a CFO, the question is whether the outcome justified the cost: Did the sales agent lift revenue? Did the customer service agent solve tickets successfully? Did it protect EBITDA? That cannot be answered from one side alone.
Treat AI Agents Like Hires, Not Software Licenses
Botanu's founders argue that an AI agent should be treated less like a software license and more like a hire. An AI agent is a new kind of workforce, and it works at 100 times the frequency of a person. 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?"
"Activity is not outcome," said Jacob. "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."
How Botanu Reconstructs Agent ROI
To answer what each agent costs and what it delivered, Botanu reads telemetry—a systems-level record of activity across a company—to reconstruct an agent's full digital footprint. It tracks across every model vendor, tool, and infrastructure layer, not just tokens. Unlike competitors, it plugs into the systems a company already runs.
It then ties that footprint to where outcomes actually land, such as the CRM, and compares it to the company's own labor data—what the same job would cost a person to do. The outcome comes from the business system that owns it, not from what the agent reports about itself.
"We showed enterprises the granular data we're able to capture over the past few weeks, and they asked us how we got it," said Alina Vrsaljko, Botanu's co-founder and CEO. "These were sophisticated technologists, and even they had never seen their agents mapped end to end."
"Token-Maxxing" vs. "Outcome-Maxxing"
Most tools measure activity—tokens consumed, calls made—and they're built for engineers. That's "token-maxxing," and it's easy to game. Botanu measures outcomes, for the business leaders who now own AI. The company calls it "outcome-maxxing." The goal is to produce more results, not more activity.
A sales agent's job isn't to make calls. It's to create qualified leads.
"Most tools look only at cost and call it ROI, or look only at value and call it impact," Jacob said. "We refuse to choose. We measure both—agent cost and business outcome—in the same view. That's the only way to answer the CFO's question."
The Measurement Problem CFOs Recognize
"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," said M.G. Thibault, who leads the Coterie CFO community and is CFO-in-residence at Scale Venture Partners. "That's the open space right now."
For executives deploying agents, the blind spot is becoming urgent. "As we start running AI agents in production, proving the ROI of each one becomes immensely challenging," said Gurpreet Bal, CIO at BHI.
Botanu's answer: tie agent performance to the same labor economics CFOs already understand. If a sales agent costs $5,000 in infrastructure and API calls per month but generates 50 qualified leads worth $10,000 each in pipeline, the ROI is clear. If it generates 3 leads worth $200 each, it's not.
The metric isn't tokens per dollar. It's outcomes per cost, measured in the business system that owns the result.
Why This Matters Now
The boardroom question has flipped. For two years, the enterprise AI story was about adoption: how many tools a company could pilot and how fast it could roll them out. In 2026, the question is what you got back. Most companies can't say.
"This isn't a bubble. It's a measurement problem," Vrsaljko said. "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."
The gap between AI spending and AI measurement is creating "AI sticker shock" in budget meetings. Botanu's thesis is that the panic is misplaced. The problem isn't the bill. It's not knowing whether you got $3 of value for every $1 you spent.
What CFOs Should Ask
If your organization is deploying AI agents in production, ask these questions:
-
Can you see the full cost of one agent? Not just API calls—infrastructure, tool usage, model vendor costs, and integration overhead across every system it touches.
-
Can you tie that cost to a business outcome? Not tokens consumed, but qualified leads created, tickets resolved, revenue protected, or EBITDA saved. The outcome should come from your CRM, ticketing system, or ERP—not from what the agent reports about itself.
-
Can you compare agent cost to labor cost? What would it cost a person to do the same job? Is the agent 10x cheaper, 2x cheaper, or 3x more expensive? Without that comparison, you can't tell if the agent economics make sense.
-
Can you performance-manage an agent like a hire? If an agent isn't delivering, can you see where it's failing, retrain it, or retire it? Or does it just keep running until someone notices the bill?
-
Can your CFO report agent ROI to the board? Not usage stats. A real KPI that ties cost to value—the one number that justifies the spend.
If you can't answer yes to all five, you have a measurement problem, not an AI problem.
The Bottom Line
Enterprises spend $186 million annually on AI agents, but most CFOs can't prove the return. The measurement gap is creating budget panic and boardroom skepticism. Botanu's "outcome-maxxing" platform ties agent costs to business outcomes CFOs recognize—qualified leads, tickets resolved, revenue protected, EBITDA saved.
The question isn't whether AI works. It's whether you can see where it's working.
For CFOs, CIOs, and COOs deploying agents in production, the discipline that's missing is simple to say and hard to do: measure outcomes, not activity, and connect costs to returns. Until then, AI is a compensation-scale expense that demands CFO-level governance most companies don't yet have.
The fix isn't spending less. It's seeing where AI actually creates value.
Sources
- Botanu Emerges from Stealth - Globe Newswire, June 11, 2026
- KPMG AI Pulse Report - KPMG, 2026
- Deloitte State of AI in the Enterprise - Deloitte, 2026
- BCG CEO AI Leadership Survey - BCG, 2026
Continue Reading
- Dell AI Factory ROI: 4,000+ Deployments Reveal Real Enterprise Costs
- [Anthropic usage-based pricing 2026: What CFOs Need to Know](/article/anthropic-usage-based-pricing-2026)
- GPT-5.4 vs Claude Opus 4.6: Performance Benchmarks for Enterprise
