Specialists, generalists, and the agentic shift
Topic:
Craft
Year:
28 January 2026
iche skill still compounds. Type design, motion systems, editorial sequencing, and identity craft are learned through repetition next to someone who already has the judgment. Studios can offer that apprenticeship in a way in-house teams and platform tools cannot, because the work rotates through enough contexts to build real range within a discipline.
What has changed is the floor. Agentic tooling now covers a large share of production tasks that used to define junior work, which means the value of a designer sits in framing the problem, judging the output, and connecting decisions across strategy, writing, interface, and motion.
The T-shaped team
The old debate framed it as a choice: hire specialists for depth or generalists for range. Agentic tools quietly dissolve that choice. When a designer can spin up a working prototype, a strategist can query a dataset, and a writer can ship production copy, the valuable person is the one with real depth in one craft and enough range to direct the rest. Depth is still the moat; range is now the multiplier on top of it.
Deep in one, literate in many
The practical model is a deep specialism paired with working literacy across adjacent disciplines. Specialism sets the quality ceiling. Literacy is what lets someone direct tooling instead of being replaced by it, and it is why interdisciplinary teams are outperforming narrow ones on the same brief.
What the tools do not replace
Agents are extraordinary at producing options and mediocre at deciding between them. They generate ten directions in the time it used to take to sketch one, which means the scarce skill is no longer production, it is taste, judgment, and the confidence to kill nine good things in service of one right thing. That is a specialist’s instinct applied at a generalist’s speed.
For studios, the implication is practical. Teams get smaller and more senior. The junior work that once justified large rosters, resizing, versioning, first-draft production, is increasingly handled by tools under supervision. What remains is the work that requires a point of view, and a point of view cannot be prompted into existence. It is earned across years of shipped decisions.
So the honest advice is neither “learn one thing forever” nor “learn everything shallowly.” Go deep enough in a craft to have standards, then use agentic tools to reach past the edges of that craft without waiting for a hand-off. The people who thrive will be unmistakably good at something and dangerously capable at the adjacent things.
N
Selected work
Other articles
S.1
Have a project like this in mind?
Tell us what you’re building. We reply to every serious inquiry within two business days.




