Botanu emerges from stealth with platform to manage AI agent ROI, revealing enterprises spend $186M annually on AI with little proof of return.
The AMW Read
Botanu is a new entrant to the AI agent segment, but the event does not involve a funding round or case-study company; it updates the agent observability/ROI measurement sub-segment with a concrete product, which is meaningful but not debate-resolving.
Botanu emerges from stealth with platform to manage AI agent ROI, revealing enterprises spend $186M annually on AI with little proof of return.
Botanu, a New York- and San Francisco-based startup founded by former McKinsey AI strategist Alina Vrsaljko and longtime enterprise AI engineer Deborah Jacob, today emerged from stealth with a platform designed to measure and optimize the return on AI agent investments. The company positions itself as "the COO for AI agents," reading telemetry across model vendors, tools, and infrastructure layers to reconstruct an agent's full digital footprint and tie that to business outcomes in systems like CRM. The launch comes as enterprise AI spending reaches an average of $186 million annually, yet a KPMG report finds only 8% of enterprises have achieved meaningful business returns with AI, and Deloitte reports that 74% of organizations want AI to grow revenue but only 20% have seen it happen.
Why it matters: Botanu's emergence signals a maturation point in the AI agent substrate. We have seen the market race through the adoption phase — piloting tools and rolling out agents — and is now entering what the founders call the "outcome-maxxing" phase, where boardrooms demand proof of return. This fits a recurring pattern: the fastest-ARR-ramp arc gave way to a capital-compression cycle, and now the industry faces the measurement and governance question that every transformative technology wave eventually confronts. Botanu's framing — treat an AI agent like a hire, not a software license — updates the open debate about whether agents are a cost center or profit center. The company's approach, reading telemetry across fragmented vendor stacks and tying spend to actual outcomes, directly addresses the problem that "adoption is not value creation," as Benchmarkit CEO Ray Rike notes. If successful, Botanu could become the performance-management layer that the enterprise AI stack has lacked, bridging the chasm between token consumption and CFO approval.
The founders make a critical distinction: the old cloud-era playbook of mapping cost to workload breaks with AI because the same task can produce wildly different costs run-to-run, and pricing has shifted from per-seat subscriptions to volatile usage-based models. "Companies aren't failing because AI doesn't work. They're failing because they can't locate where their agents are working," said Vrsaljko. Independent analysts concur. "Token spend is up 13x since January 2025, yet only 27% of executives say AI has met their ROI expectations," said Rike. The discipline that's missing, he argues, is measuring outcomes rather than activity and connecting costs to returns. Botanu's bet is that enterprise AI is not overfunded but undermeasured, and that the fix for "AI sticker shock" isn't spending less but seeing where AI actually creates value.