The $2 Billion Question No One Could Answer
The CFO of a Fortune 500 diversified industrial company sat across from her transformation team with what seemed like a straightforward request. The board wanted to accelerate capital allocation decisions to stay competitive with private equity firms circling their industry. “How long,” she asked, “does it take us to go from receiving an investment proposal to delivering a board-ready recommendation?”
The silence that followed wasn’t comfortable.
“It depends,” the VP of Capital Planning finally ventured.
“On the business unit, the proposal type, the time of year, who’s available...”
“Give me a range,” the CFO pressed.
“Anywhere from four weeks to four months.”
The CFO looked at the scattered nods around the table. Different people, different answers. For a function that allocated $2 billion in capital annually—decisions that literally shaped the company's future—no one could articulate with precision how those decisions actually got made.
This isn't an unusual situation. It's the norm.
And it reveals why most agentic AI transformations in strategic functions like capital allocation fail before they begin. You can't automate what you haven't defined. You can't orchestrate what you don't understand. And you can't deploy intelligent agents into a process that exists primarily in people's heads.
The Reality Check: Where AI Transformation Stalls
The enterprise AI market is on fire. According to PwC’s 2025 AI Agent Survey, 79% of organizations have adopted AI agents at some level, and 88% plan to increase AI-related budgets in the next 12 months. Gartner predicts that by 2028, 33% of enterprise software applications will include agentic AI, up from less than 1% in 2024.
But here’s the uncomfortable truth hidden in those optimistic numbers: McKinsey’s 2025 State of AI report reveals that while 23% of organizations are scaling agentic AI systems somewhere in their enterprise, most are stuck in experimentation. Nearly two-thirds haven’t begun scaling AI across the enterprise. Only 39% report EBIT impact at the enterprise level.
The gap between pilots and production is massive.
And for strategic functions like capital allocation—where decisions involve judgment, nuance, and high-stakes trade-offs—that gap is even wider. A 2021 EY survey of 1,050 CFOs found that 56% said their capital allocation strategy needs to be completely rethought, and 80% believed their capital allocation process needs improvement. Yet here we are in 2026, and most are still operating on spreadsheets, tribal knowledge, and manual workflows.
Why? Because they’re approaching the problem backwards.
The Fatal Assumption: "Let's Just Apply AI to Our Process"
Technology vendors make it sound easy: “Deploy our agentic AI platform and watch your capital allocation transform!” Systems integrators promise: “We’ll implement AI agents in 12 weeks!”
What they don’t tell you is that their success depends entirely on something they can’t provide: clarity about how your process actually works.
Most organizations operate under a dangerous assumption: “Our capital allocation process is basically fine—it’s just slow and manual. If we automate it with AI, we’ll get the same outcomes faster.”
This assumption is wrong on multiple levels.
First, if you can’t measure your current state, you can’t know if AI is making it better. When JPMorgan suffered a $6.2 billion trading loss partly due to a spreadsheet error in their risk model, it wasn’t a technology problem—it was a process visibility problem. When enterprises deploy capital to suboptimal investments because they’re evaluating projects sequentially rather than holistically, that’s not an efficiency problem—it’s a design problem.
Second, strategic functions like capital allocation are fundamentally different from operational processes where RPA and traditional automation thrive:
- Operational processes are repetitive, rule-based, clearly defined, and stable over time. Think invoice processing: same 12 fields, same approval hierarchy, 10,000 times per month. You can map the happy path, automate the clicks, handle exceptions.
- Strategic processes are variable, judgment-intensive, context-dependent, and evolving with market conditions. Every capital investment is unique. Assumptions are debatable. Trade-offs are subjective. The “right answer” depends on risk appetite, strategic priorities, and factors that won’t appear in any financial model.
Applying operational automation playbooks to strategic functions is like using a hammer when you need a scalpel.
The Fatal Assumption: "Let's Just Apply AI to Our Process"
Technology vendors make it sound easy: “Deploy our agentic AI platform and watch your capital allocation transform!” Systems integrators promise: “We’ll implement AI agents in 12 weeks!”
What they don’t tell you is that their success depends entirely on something they can’t provide: clarity about how your process actually works.
Most organizations operate under a dangerous assumption: “Our capital allocation process is basically fine—it’s just slow and manual. If we automate it with AI, we’ll get the same outcomes faster.”
This assumption is wrong on multiple levels.
First, if you can’t measure your current state, you can’t know if AI is making it better. When JPMorgan suffered a $6.2 billion trading loss partly due to a spreadsheet error in their risk model, it wasn’t a technology problem—it was a process visibility problem. When enterprises deploy capital to suboptimal investments because they’re evaluating projects sequentially rather than holistically, that’s not an efficiency problem—it’s a design problem.
Second, strategic functions like capital allocation are fundamentally different from operational processes where RPA and traditional automation thrive:
- Operational processes are repetitive, rule-based, clearly defined, and stable over time. Think invoice processing: same 12 fields, same approval hierarchy, 10,000 times per month. You can map the happy path, automate the clicks, handle exceptions.
- Strategic processes are variable, judgment-intensive, context-dependent, and evolving with market conditions. Every capital investment is unique. Assumptions are debatable. Trade-offs are subjective. The “right answer” depends on risk appetite, strategic priorities, and factors that won’t appear in any financial model.
Applying operational automation playbooks to strategic functions is like using a hammer when you need a scalpel.
