A Two-Speed Problem That Used to Be a One-Speed Problem
Until recently, AI maturity was something you compared agency to agency. One shop had it together. The one down the street didn't.
That's changed.
Spark AI's 2026 report found the gap between the most active and least active AI users inside a single agency is now wider than the gap between one agency and another. Inside one team: 45% of people use AI daily. 40% dip in and out with no consistency. 15% barely touch it at all.
That's a two-speed workforce. Under one roof. Working on the same client projects.
You cannot standardise quality around that. You cannot guarantee consistency. You cannot build repeatable processes on top of it. Everything your client expects to be uniform — tone, craft, turnaround — is sitting on a workforce that's operating at radically different capability levels.
The Middle 40% Decides Your Future
Most leadership conversations about this go to the edges. "We need to get the 15% engaged." Or: "We need to unleash our 45%."
Both are wrong.
The 15% is a long and slow project. Some of them never engage — that's a different conversation. The 45% are already moving. They don't need you for acceleration, and often the structure you'd add slows them down.
The 40% is the group that decides whether your agency advances or stays where it is.
They're not resistant. They're not enthusiasts. They're competent people with too many client fires to fight, and the evidence they've seen so far hasn't been enough to change how they work. They need one clear first step, not a strategy deck.
Why the Enthusiasts Alone Won't Save You
Every agency I work with has a story like this. "We've got two or three people who are incredible with AI." And they do. Those people are real.
They're also fragile.
Spark's data shows that when the middle of the team doesn't move, the enthusiasts burn out or leave. They read the absence of organisational investment as a sign that the work they're doing doesn't matter to leadership. And they're not wrong.
An agency that relies on three people for its AI capability has three points of failure, not three points of strength. If any of them walk, the workflows walk with them. That's now showing up in M&A due diligence — buyers look for systematised capability, not individual heroes.
The Buddy System That Actually Moves the Middle
Here's what works, tactically.
Pair one daily user with one inconsistent user. Two weeks. One specific assignment: build one repeatable workflow for the inconsistent user's most repetitive weekly task.
Not "use more AI." Not "get better at prompting." One workflow. One weekly task. Documented by the end of two weeks.
The daily user gets the validation of teaching. The inconsistent user gets over the first-move barrier that was keeping them inconsistent. And your agency gets an SOP it didn't have before.
Do this across five pairs simultaneously and by the end of the month you have five new shared workflows, ten people who worked together on AI for the first time, and early visibility into who's actually capable versus who's been claiming to be.
That last one matters. Spark's data is blunt: 83% of agency staff self-report as capable AI users. Only 15% actually operate at that level. The buddy system surfaces who's where without making anyone defend themselves in a meeting.
Stop Guessing. Get the Dashboard.
The reason most teams are stuck in the two-speed mode is that leadership has no visibility. When every user has a personal AI account, there's no picture of what's happening. Who's using what. How often. For what.
Move to team-level subscriptions. Shared workspaces. A dashboard leadership can actually look at. Once you can see the shape of the gap, closing it becomes a management problem instead of a mystery.
The agencies pulling ahead are treating AI adoption as standing leadership business — department reviews, performance conversations, M&A preparation. Not as a cultural side project.
The gap is real. It's measurable. And it's closable. But only if someone in charge decides to close it.