Hiring an "AI lead" is the wrong response to this moment. The teams that benefit are the ones where every craft — strategy, design, copy, eng — uses models in the work itself, not as a separate layer bolted on top.
The pattern we see: agency leadership reads three articles, panics gently, and creates a "Head of AI" role. Six months later, that person is producing decks instead of moving the practice forward. The org didn't change shape — it just acquired a new title.
What works instead
The teams making real progress treat AI like a horizontal capability — closer to typography or interaction design than to a department. Every craft uses it. The models become part of the toolchain, not a stop on the workflow.
Three things follow from that:
- Hiring shifts. Senior strategists, designers, and engineers who use models confidently are worth more than juniors plus an "AI specialist."
- Pricing shifts. Faster output means hourly billing breaks. Outcome-based pricing — or fixed-scope deliverables — fits better.
- Process shifts. Reviews happen earlier and more often, because the cost of producing the next iteration is small.
“AI is to this decade what the spreadsheet was to the 80s. The CFO didn't get an AI lead in 1985 — they hired accountants who could use Lotus.”
The agency operating system
What we've built internally is closer to a set of habits than a tech stack. Models in every craft, weekly demos, paired reviews, and an internal log of prompts that worked. None of it requires a department.
If you're an agency lead trying to figure out the right move: don't hire an AI specialist. Pick three projects, mandate that every team member uses models in their part of the brief, and review the output together at the end of the week. Repeat for a quarter.