The Untapped Lever in Your AI Adoption Strategy

Most AI adoption strategies for engineering organizations focus on the same group: the individual contributors. The developers who need convincing, the early adopters pulling ahead, the skeptics in the middle who aren't moving. That focus is warranted — but it's incomplete. There's another group sitting at the center of every engineering team that almost no adoption strategy addresses directly, and new data suggests they may be the highest-leverage variable most leaders aren't using.

Engineering managers. Not as a problem to solve, but as an untapped engine — one that, when engaged, has an outsized effect on both team adoption and direct technical contribution that few organizations have started to fully exploit.

According to DX's Q4 2025 AI Impact Report, which analyzed pull request data from 385 engineering managers, those using AI daily were shipping twice as many PRs as those who rarely used the tools. By Q1 2026, that figure had doubled again — according to DX's Q1 2026 Impact Report, engineering managers using AI are now shipping four times as much code as their non-adopting peers. That is a remarkable finding on its own. But the more important implication isn't about the code. It's about what happens to the teams underneath those managers.

That 4x figure is striking. But spend a moment with the second-order implication: if engaged managers are contributing at that level while simultaneously influencing the adoption behavior of every developer on their team, the compounding effect is enormous. The manager isn't just a contributor — they're a signal, a model, and a cultural driver all at once. Ignoring that leverage isn't just a missed opportunity. It's one of the most common reasons adoption programs stall despite significant investment.

Why Managers Are Uniquely Positioned — and Usually Overlooked

For years, the conventional wisdom on engineering management has been clear: once you move into management, your job is people, process, and priorities. Writing code is optional at best, a distraction at worst. And structurally, that's exactly what happens. As engineers get promoted, their calendar fills with standups, roadmap reviews, one-on-ones, and planning sessions. The coding time disappears — not by choice, but by gravity. By the time someone is a senior engineering manager, their last meaningful commit might be years behind them.

AI agents are changing that equation in a way nothing else has. The barrier to switching between strategic and technical work has collapsed. Agentic tools can take a clear description of intent and execute work that used to require hours of uninterrupted coding focus. For managers whose time is sliced into thirty-minute windows, this is genuinely transformative. You don't need a full afternoon to contribute meaningfully to the codebase anymore — you need judgment, clear communication, and the ability to evaluate what comes back.

This is not a coincidence. The managers who became excellent at leading developers built those skills through years of translating ambiguity into clarity, challenging technical approaches, and knowing when a solution was right or wrong without writing every line themselves. Working effectively with an AI agent requires precisely the same toolkit. The manager who learned to extract the best work from a team of developers is already equipped to extract the best work from an agentic workflow. They just haven't been shown the connection yet.

AI also gives managers something they haven't had in years: an avenue to complete meaningful technical work autonomously, without competing for developer time or adding to someone else's queue. That's a significant shift. It means a manager can prototype an idea, validate a technical assumption, or complete a well-scoped task without pulling a developer off their priorities. As DX researchers noted in their Q1 2026 report, this is enabling the return of the true player-coach — the engineering manager who leads and still contributes. The question is which of your managers is ready to play that role.

The Team Effect: Why Manager Adoption Spreads Differently

Individual developer adoption spreads horizontally — peer to peer, team to team, mostly by word of mouth and shared tools. Manager adoption spreads vertically, and it moves faster. When a manager is visibly engaged with AI — using it, talking about it authentically, demonstrating what it makes possible — that signal reaches every developer on their team simultaneously. It changes the social permission structure around trying new things. It shifts the question from "is this allowed?" to "how do I get started?"

The inverse is equally true, and equally powerful. A manager who is skeptical, disengaged, or simply absent from the conversation about AI tools sends a signal too — and developers read it clearly. If the person responsible for their performance review isn't prioritizing this, why should they? This is the silent multiplier that most AI adoption strategies never account for, and at Lever10 it is one of the most reliable predictors we track when assessing why well-resourced programs fail to gain traction.

The data from DX makes this concrete: the managers who aren't using the tools are the ones whose teams have the lowest adoption rates. That pattern is not subtle. And crucially, it means that winning over a single manager has a cascade effect that winning over a single developer simply doesn't.

This reframes where leaders should be spending their energy. If you have limited bandwidth to drive adoption, the manager layer is where that investment compounds the most. One engaged manager, genuinely modeling the behavior, is worth more to your adoption numbers than a dozen individual developer nudges.

How to Use This — Practically

The answer is not to start measuring managers' PR counts. That misses the point entirely and creates the wrong incentives. The goal is genuine technical re-engagement — managers who are in the tools because they see the value, not because someone is watching.

Start by sharing this data directly with your managers. Show them the 4x figure. Share this article. Make the expectation explicit: this is what AI-engaged engineering leadership looks like, and it is what is now expected. Most managers who have drifted from the code did so because the organizational signals told them to — they were rewarded for meetings, roadmaps, and headcount decisions, not for shipping. If you want them back in the tools, you have to say so, protect the time, and normalize the process of being a beginner again at something.

Then invest in their development directly. The same managers who resist AI engagement often carry the same underlying fear as the mid-level developers on their teams — that the skills they spent years building are becoming less relevant. They need the same reframe: their judgment, their ability to communicate intent clearly, their instinct for evaluating architecture and asking hard questions — those skills don't become obsolete with AI. They become the core competency that separates managers who use these tools powerfully from those who can't.

The managers who make this transition become your most valuable force multipliers — contributing code directly while simultaneously pulling their teams forward. That dual impact is rare and compounding. A manager who uses AI agents daily, who talks about it authentically with their team, who can sit alongside a skeptical developer and show them what good looks like — that person accelerates your adoption goals faster than any program you can build around them.

That is not a comfortable thing to say. But the leaders who act on it early — who replace the managers who won't make the transition with ones who will — are the organizations that will look back on this period as the moment they pulled decisively ahead. The ones who wait will wonder what happened.