The Great Inversion

Edison said genius is 1% inspiration and 99% perspiration.

For a century, that ratio shaped how we built everything. Org charts. Hiring criteria. Promotion paths. Consulting models. Project management. All of it optimized for the 99%. Because the 99% was expensive.

Last week at Anthropic's Global Partner Summit, Eric Burns reframed it in a way I haven't been able to stop thinking about: that ratio is flipping. Fast. Execution is getting exponentially cheaper.

This isn't about AI getting smarter. It's about your company still being built for expensive execution.

TL;DR:

• Edison's ratio is flipping: genius is shifting toward 99% inspiration, 1% perspiration
• On agentic benchmarks, AI handles tasks that took humans 14+ hours, up from 2 minutes in 2023 (METR data).
• Most organizations were structured to coordinate expensive execution. That infrastructure needs a new purpose now that execution is getting cheap.
• The skill that matters now: knowing which problem to solve. Most organizations never systematically developed that below leadership.

The Old Ratio

Think about what the 99% perspiration world built.

We hired executors. The best engineer, the fastest analyst, the most reliable project manager. We promoted them into leadership because they were the best at doing the work. We structured organizations with layers of management whose primary job was coordinating execution across teams. We priced services by the hour because execution was the scarce resource.

Every status meeting, every approval chain, every QA gate, every project management ritual exists because execution used to be expensive, slow, and error-prone. Coordinating it required infrastructure. That infrastructure became the organization.

This wasn't wrong. When building a feature took a team of five engineers three months and $200K, you needed project managers to track progress, QA to verify quality, managers to resolve dependencies, and directors to prioritize what got built at all. The coordination cost was justified by the execution cost.

The Flip

METR, an independent AI evaluation lab, has been measuring this: the complexity of tasks AI can handle autonomously is roughly doubling every seven months. Not in theory. In measurement.

In 2023, AI could reliably complete tasks that took a human under two minutes. As of early 2026, that threshold crossed fourteen hours. Same measurement framework, same human baseline. The curve is nearly vertical.

In practice: industries with the highest compliance burdens are compressing work that took teams months into hours. A software project that would normally occupy a senior engineering team for months was built by a team of parallel AI agents for $20,000 in compute (Anthropic, February 2026).

These aren't demos. They're in production.

The result is a gap that's widening faster than most leaders realize. Model capabilities are accelerating exponentially. Business capabilities are growing linearly, if at all. The space between those two curves is the most important number in enterprise technology right now. Not your AI budget. Not your model selection. The gap between what AI can do and what your organization can absorb.

The Coordination Layer

Here's where it gets uncomfortable.

Every organization I work with has what I've started calling the Coordination Layer: the management infrastructure built to coordinate expensive execution. Status meetings. Approval chains. Project management offices. QA gates. Reporting hierarchies. Resource allocation committees.

The Coordination Layer made sense when execution took months and cost hundreds of thousands of dollars. You needed people whose full-time job was making sure execution happened correctly, on time, and on budget.

But execution is collapsing in cost and duration. And the Coordination Layer hasn't been told to change.

I worked with a mid-market services firm last year. 200 people. They had three layers of management between the CEO and the people doing client work. When I asked what those managers spent their time on, the answer was consistent: coordinating. Scheduling. Tracking. Syncing. Resolving dependencies between teams.

We ran a pilot automating one client workflow. The work that used to take a team of four people two weeks finished in an afternoon. The output was equivalent. The four people found more valuable work to do.

But the three managers who coordinated that work? They were still scheduling standups. Still sending status updates. Still running the weekly sync meeting for a process that now took hours instead of weeks.

Nobody told them the ratio is flipping.

If you're reading this and recognizing your own role, you're not the problem. You're the person who knows how work actually gets done in your organization. That knowledge is exactly what the 1% needs. The question is whether your job description catches up to what you're actually capable of.

What the 1% Actually Looks Like

The 1% that remains is the hardest part to hire for.

Problem identification. Scoping. Judgment. The decision to solve this problem and not that one. The conviction to kill a project that's working but isn't essential. The ability to look at an organization and see where the real friction lives, not where it's most visible.

These skills were reserved for the executive layer. Never scaled down. Never the reason a mid-career employee got promoted.

But now every role is becoming a judgment role. And nobody built the training, the incentives, or the career paths for that.

We spent fifty years promoting the best executors into leadership positions. The fastest coder became the engineering manager. The most productive analyst became the finance director. The most reliable project manager became the VP of operations.

Now execution is a commodity. And the people running organizations were selected for a skill that's being commoditized.

This isn't a training problem. It's closer to what I wrote about in The Organizational 95%: an identity crisis. You're asking leaders who built their careers on execution excellence to redefine their value around judgment and problem identification. Skills they may have, but were never recognized for.

The organizations I see getting this right share a pattern. They invest disproportionately in the 1%. Long discovery phases before any technology is deployed. Deep stakeholder interviews. Process archaeology that maps how work actually happens, not how the documentation says it happens.

The 70-20-10 rule (BCG research) takes on new meaning here. 70% people and process, 20% technology, 10% algorithms. But now the 70% isn't about training people to use AI. It's about training people to think about which problems are worth solving. That's the inspiration. The rest is perspiration the machines handle.

The New Shape

Here's the thing: most organizations aren't failing at AI because they chose the wrong model or the wrong vendor. They're failing because they're still shaped for the old ratio.

An organization built for the old ratio looks like a pyramid. Small strategic layer at the top. Massive execution layer below. Layers of coordination in between. The value was in the base.

The new ratio inverts the pyramid. The execution layer, now AI-powered, can be vastly larger for a fraction of the cost. The coordination layer shrinks because there's less to coordinate. The strategic layer, the 1% at the top, becomes the bottleneck and the differentiator.

If your organization still looks like the old pyramid, you're carrying infrastructure that was designed for expensive execution. Every coordination meeting, every approval chain, every reporting layer that exists to manage execution was justified by the old cost structure. That cost structure is changing.

This isn't about cutting headcount. The people in those roles aren't the problem. The job descriptions are. The managers coordinating execution can be redirected toward the work that actually creates value now: problem identification, customer insight, strategic prioritization, quality judgment.

But only if someone names the shift.

The mid-market firms that have named it are getting it right. Fewer coordination layers. Longer discovery phases. A bias toward killing projects that aren't essential. They spend most of their AI investment on the 1%, not the 99%. They look different. And they're pulling ahead.

Your Move

Two steps. One this week, one this month.

This week, ask your leadership team one question: if AI handled all the execution tomorrow, what would we still need humans for?

If the room goes quiet, that's your answer. The organization doesn't know what its 1% is. Everything downstream of that, your AI strategy, your hiring plan, your org design, is built on a foundation that doesn't exist yet.

This month, pick five roles on your org chart. For each one, ask: does this role exist primarily to coordinate execution? If the answer is yes for three or more, your organization is still shaped for the old ratio.

That doesn't mean those roles disappear. It means they transform. From coordinating execution to curating judgment. From managing output to identifying which problems deserve investment. From the 99% to the 1%.

The ratio is flipping. Your org chart isn't.

Start there.