There is a Friday afternoon most senior leaders eventually recognize.

You close the laptop, look back at the week behind you, and notice something you have been avoiding.

The three or four strategic questions that only you can answer, the directional calls your function actually needs from you, have not moved.

The reflex is to read this as a discipline problem. Protect time better. Push delegation harder. Get more ruthless about the calendar.

That reflex is wrong. Most of the time, the discipline is a symptom. The structure is the cause.

When I was running a $600M product portfolio at GoDaddy in 2017, my calendar that quarter held seven direct reports with weekly business reviews, four cross-functional steering committees, and the executive forums that came with sitting at the table where capital decisions got made.

The architecture of the job optimized for visibility and coordination.

The thing the company actually needed from me, executive judgment on the small number of decisions only I could make, had no protected slot. It was whatever was left over after the operational reviews finished, which is to say, almost nothing.

I thought about that pattern again when I read the recent reporting on Dario Amodei. The Anthropic CEO has exactly one direct report, his chief of staff.

His sister and co-founder Daniela serves as President.

Every other executive, including the CFO and CCO, reports through her. In a June 2026 Bloomberg interview, Amodei described the separation between strategic long-term thinking and short-term operations as "incredibly freeing."

Read the full essay on Substack for the raw, personal reflection on how this calendar trap played out during my time at GoDaddy.

The structure runs counter to almost everything Silicon Valley currently rewards. Jensen Huang manages roughly 60 direct reports at Nvidia. Sam Altman has about half a dozen at OpenAI.

The conventional read is that more reports means more leverage.

Amodei has gone the other direction, deliberately. He spends up to 40% of his time on culture, with the remainder on AI research direction and the long-form policy essays that have become part of how Anthropic shapes the broader conversation.

Daniela runs the organization. He shapes the ideas.

The question underneath the structural choice is the one most senior leaders refuse to ask honestly. What is the small number of things only I can do, and what would it take to protect them?

In nearly every coaching engagement I have run since 2021, the senior leader cannot answer the first half with any precision.

They have not separated the work that requires their judgment from the work that requires their presence. The two get conflated, and presence becomes the proxy for contribution.

A CTO at one of the large platform companies told me he had fourteen direct reports and could not name a single strategic decision he had personally moved that quarter.

A VP at Microsoft told me she had started measuring her own importance by how full her calendar was.

A senior leader at Apple admitted, three months into our work together, that he had not had a clean two-hour thinking block since taking the role.

The conflation gets more expensive, not less, as the role gets more consequential. In the AI era, the marginal cost of a senior leader spending a Tuesday on a budget reconciliation someone else could run is no longer measured in hours.

It is measured in the strategic frames that did not get set, the directional calls that did not get made, the bets that defaulted to whatever was easiest to schedule.

The metric most senior leaders still optimize is span of control. The metric that actually matters in AI-era leadership is span of attention. The two are often inversely correlated.

Span of control reflects how much of the organization runs through you. Span of attention reflects how much of your judgment makes it into the decisions that matter.

AI is compressing the value of coordination work faster than most leaders are repricing their calendars to match.

The translation roles, the synthesis-and-route work, the meetings that exist to align other meetings, those compress first. What remains is the small number of decisions where executive judgment is still the bottleneck.

Most senior calendars were designed for the previous regime.

Looking back at my 2017 calendar, the part I find hardest to forgive is not that the strategic questions went unanswered. It is that I had assumed the calendar was a reflection of how important my work was. It was nothing of the kind. It was a record of which work was easiest to schedule.

If you find yourself reading this on a Friday afternoon, the test is not whether you can list more things to delegate. It is whether you can name the three or four decisions over the next quarter that would actually require you, and whether any of them has a protected slot.

If the answer to the second is no, the problem is not your discipline. It is your architecture.

The argument above is the structured framework version. If you want the fuller narrative take on the same question, including where the Dario Amodei observation comes from and what it looked like from the inside, I published a longer piece on Substack this week: maheshmthakur.substack.com. New essays on executive leadership and AI, twice a week.