Determined Generalist

When the boundary of a craft moves

For roughly the past century, work meant having a craft or a subject-matter expertise and deferring the moment you hit its edge: "you're a great engineer, but you enter design, so ask the designer"; "you have ideas, but go ask someone skilled at sales." Skills were scarce, so organizations were assembled from specialists who handed work across boundaries. Eric Glyman describes a principle for how large language models change that arrangement. In his telling, LLMs invert the scarcity of expertise: they have "read more lines of code than any engineer alive, more medical case texts than any doctor alive, more company filings and charts of accounts than any accountant who's ever lived or ever will."¹

The consequence he draws is personal and immediate. "If I can just ask a good question, I'm the best doctor I've ever been in my life, and I'm not a doctor at all. If I'm just stubborn, I can keep going."¹ The binding input, in his framing, shifts from having the skill to having the determination and the willingness to build systems. A stubborn generalist can now push past their old boundary rather than stopping at it.

Two consequences Glyman draws

The first is a hiring consequence. Glyman argues the aperture widens: "if you find really determined people with a high level of determination and an aptitude, just a drive, these people can do far more than they ever used to."¹ The value is no longer restricted to people carrying a specific rare spike; determination becomes a first-class thing to select for. He treats this as a complement to, not a replacement for, the practice of hiring for extraordinary spikes, which he and Karim Atiyeh also run at Ramp. Returns to the very best in a narrow area remain high; the general aperture simply widens to include determined generalists as well.

The second is an org-design consequence. Glyman observes that organizations built out of many sub-specialties create "walls and barriers and things that get lost between organizations." He describes a pattern he has watched inside Ramp: at around ten people someone says "I work at Ramp and I'm an engineer," but by two hundred to five hundred the identity inverts to "I'm a salesperson at Ramp, I'm a designer at Ramp." People begin to identify by craft and interact mostly within it, no one in sales knows an actual engineer, and everything routes up and down through the hierarchy. When a single person who builds the product can also use tools to sell, track customers, and write a serviceable pitch on the first attempt, he says, "you start seeing fewer types of specialties within organizations," and the org's shape can be radically simplified.

How it connects and where it strains

Glyman frames this as the talent-and-org consequence of knowledge becoming cheap to summon: when the marginal cost of knowledge falls, the scarce input becomes the drive to keep going and the judgment to ask the right question. It reframes hiring, org design, and the reason hierarchies exist at all. It sits alongside the A-player framework as another lens on who is irreplaceable, though the two select for different things, one for an irreplaceable superpower and this one for irreplaceable persistence.

Glyman himself notes the limits. The claim that a determined generalist can do more has a ceiling in genuinely deep, credentialed, or high-liability domains, and his "best doctor I've ever been" line is offered as a provocation rather than a licensing claim. Collapsing specialties can also lose the depth that true specialists provide, leaving open which functions genuinely compress under a generalist-plus-tools and which still require an expert. Determination is also hard to interview for, and detecting it tends to lean on the same proof-of-work signals used to detect spikes, which may under-select quieter high-agency people.