Every enterprise I advise makes the same initial mistake: they hand AI to their IT department and wait for magic to happen.

It rarely does.

The problem isn't technical capability. Modern LLMs can draft contracts, summarize earnings calls, triage support tickets, and generate code reviews. The problem is organizational framing. When you treat AI as "software," you get software outcomes: vendor evaluations, security reviews, pilot projects that never scale.

When you treat AI as a supply chain issue, everything changes.

The Labor Lens

Here's the mental model that's worked for my clients: LLMs are a new form of labor, not a new form of software.

Think about what your organization actually buys from knowledge workers: judgment applied to information over time. That's exactly what an LLM provides, at a fraction of the cost and infinitely scalable hours.

This isn't about replacing people. It's about recognizing that your organization has labor bottlenecks that no amount of hiring will solve. The question becomes: where is cognitive labor most constrained?

The Supply Chain Question

If you could add 100 junior analysts to any function tomorrow, at zero marginal cost, where would they create the most value?

The Framework

I use three criteria to identify high-value deployment zones:

  1. Volume: Is there enough repetitive cognitive work to justify the setup cost?
  2. Variance tolerance: Can the process handle 80% accuracy, or does it need 99.9%?
  3. Velocity pressure: Is speed a competitive advantage in this function?

Functions that score high on all three—customer support, first-pass document review, internal knowledge synthesis—are where I start. Functions that score low on variance tolerance—regulatory filings, financial statements, legal opinions—are where I wait.

What This Means For Your Organization

Stop asking "what can AI do?" and start asking "where is cognitive labor constraining our business?"

The answer usually isn't in IT. It's in operations, finance, legal, HR—the places where smart people spend their days on work that's important but not differentiated.

That's where the supply chain is broken. That's where AI becomes strategic.