Anthropic published something important this week.
Their head of economics, Peter McCrory, released a study based on real Claude usage data across enterprise accounts. The headline number: 94% of software developer work is theoretically automatable by AI.
The second number is more interesting: roughly 30% is actually being automated today.
That 64-point gap — between what AI could do and what it's actually doing — is the most important number in workforce strategy right now. And most people are reading it wrong.
Why the Gap Exists
The 94% is about capability. Claude can, in principle, handle the cognitive steps involved in the vast majority of developer tasks: code generation, debugging, documentation, code review, test writing. The underlying capability is real.
The 30% is about deployment. Organizations haven't built the workflows, governance structures, or human integration layers to capture that capability at scale. The tools are available. The architecture to use them isn't.
This is an execution problem, not a technology problem.
What It Means for Your Hiring
If 30% of developer work is already being absorbed by AI — and you're staffing as if the number is 0% — you're building the wrong team.
The companies ahead of this are already staffing for where the number is going, not where it is. They're not hiring junior developers to execute tasks that AI handles. They're hiring engineers who can design the systems that direct AI execution.
That's a different hire. It pays differently. It interviews differently. It comes from a different pool.
Most job descriptions haven't caught up. Most interview processes can't tell the difference. That gap — between what companies are hiring for and what they actually need — is where talent strategy gets expensive.
The Skills Gap Inside the Gap
The Anthropic study flagged something else: a widening divide between AI power users and everyone else inside enterprise teams. Early adopters — engineers who've spent 18 months integrating AI into daily workflows — are dramatically more productive than teammates who haven't.
This matters more than the headline number.
You probably have people in your organization operating at 50-60% personal automation already. And sitting next to them: people still doing everything manually. Same payroll. Very different output.
The hiring implication is direct: screening for AI-native work habits is now more predictive of output than screening for credentials. A developer who's spent the last year directing AI agents to write their code has a fundamentally different capability profile than one with a better resume who hasn't touched the tools.
Most ATS systems can't detect this. Most hiring managers aren't asking the questions that surface it.
The Window Closes
The 64-point gap between theoretical and actual automation is temporary. Every month, the actual number moves up. Organizations that crack the deployment problem first will be running leaner, faster teams than the ones still working on workflow architecture.
That transition creates two distinct markets: companies building AI-native teams now, and companies competing for the same talent 18 months from now when the gap has closed and the advantage is already gone.
The window to build this intentionally — rather than reactively — is open right now.
VC5 Consulting works with companies figuring out how to staff for AI-augmented operations. If you're trying to understand what your team should look like at 50% automation — and how to hire for it — let's talk.