Operate at All Levels

The one trait that only shows up in winners

When Brex began what Pedro Franceschi calls "Brex 3.0," the company ran an exercise on its own leadership: separate the most successful leaders from the least successful and identify what actually predicts the difference. Franceschi describes the methodological problem that made the exercise hard. Most traits that correlate with success also appear in the people who failed, which makes correlation nearly useless as a filter.¹ The one attribute he says showed up only among the winners was the ability to "operate at all levels," which he defines as the capacity to "switch from 40,000 feet to five in like 10 seconds."¹

The top altitude, in his account, is the mission, the strategy, the long-range plan, and the financial outcomes. The ground level is the product design, the engineering architecture, and how the thing is actually being built by an individual contributor. A leader who operates at all levels holds both at once and moves between them quickly.

The scaling failure mode it defends against

Franceschi frames the trait as a defense against a structural problem that arrives with scale. As a company grows, he argues, it "starts operating through proxies and loses connectivity to what it's actually shipping."¹ Layers of management, reporting, and abstraction accumulate between the people setting direction and the product that reaches customers. Leaders who can drop to the metal, in his telling, keep that connection alive, which is why he says Brex tries to run "much closer to the metal" than its size would suggest.

This connects the trait to two other pieces of the Brex playbook Franceschi describes. Products Ship 90% Complete depends on leaders who can judge both whether a problem is worth solving at the 40,000-foot level and whether the actual build is good at the 5-foot level. Start New Things, Don't Change the Company depends on operators who can run a small unit with founder-level attention to detail while still tying it to the strategy of a much larger organization.

The supporting traits and the baseline

Franceschi pairs operate-at-all-levels with two further attributes he says the exercise surfaced. The first is that great builders treat their work as a reflection of themselves and hold their own quality bar higher than any bar imposed from outside, so they do not need to be told to be good. The second is that intensity is a daily choice, which he summarizes with the observation that "companies choose to be fast," and a diagnostic he uses to test it: when you follow up on something, what counts as an acceptable answer, next week or later today.¹ Underneath all three, he describes a skill baseline of high technical ability and strong product taste, often in people who have built with nondeterministic technology since before ChatGPT.

Where the trait is contestable

Franceschi is explicit that the trait was found by studying Brex's own successful leaders, which introduces the risk that it describes Brex's particular culture rather than a general law. His own guard against this is the rule that traits which also appear in failures do not count, which narrows the field but does not remove the survivorship concern. A second tension he leaves open is that if altitude-flexing is precisely what preserves ground-connection, and scale is precisely what erodes ground-connection, then the trait becomes rarer exactly as a company hires quickly, making the doctrine a continuous balancing act rather than a fixed solution.