The line manager role is disappearing faster than most career plans assume. Meta has already moved toward one senior manager for roughly every fifty developers. That is not a distant forecast. It is an operating choice already in the market. If you lead engineers — or want to — the useful question is not “will this happen?” It is what kind of leader still has a job when the org chart gets that thin.
Where this is going
I see three moves landing at once:
A. AI agents write most of the code. One senior developer directing a swarm of agents is not science fiction. It is the near-term shape of delivery: humans own intent, constraints, and review; machines own volume. Span of control for output rises even when headcount does not.
B. The middle manager who mainly runs one-on-ones and tracks tickets will struggle to justify the seat. Coordination theater does not survive a world where execution bandwidth is cheap and coordination tools are better. If your value is status meetings and a dashboard of green/yellow/red, you are competing with software that does that without a skip-level.
C. The leader who thrives will look different from the leader who thrived in the last decade. Not “more soft skills” as a slogan. A different job: business outcomes with technical teeth, not people-process as the product.
Those three reinforce each other. When coding throughput jumps, the bottleneck moves to judgment — what to build, what to kill, what interface to freeze, what risk is acceptable. Judgment is leadership work. Ticket wrangling is not.
What the surviving leader actually owns
The leaders I would hire into a flatter org share a pattern:
- They own business outcomes, not just a people inventory. Headcount is an input. Revenue, reliability, adoption, and cost of delay are the scoreboard.
- They understand product, engineering, operations, and how sales actually sells. Enough to argue tradeoffs in the same language as partners — not as a translator who only ships Jira.
- They think in systems, not only sprints. Feedback loops, failure domains, platform interfaces, incentive design. Sprint hygiene still matters; it is not the strategy.
- They keep enough technical depth that returning to IC tomorrow is credible. Not day-one expert on every stack — but not a spectator in design reviews either.
The old clean split — pure management track vs pure IC track — is already leaking. Hybrid value is the default. Specialization remains; pure process ownership without domain force does not.
If I stepped into an EM role tomorrow
I would do three things immediately, and keep doing them:
- Stay technically current. Read designs. Ship small things. Touch the tools agents use. You cannot review what you cannot feel.
- Invest hard in strategic and business thinking. How the company makes money, where margin dies, which customers actually drive the roadmap. Engineering leadership without that is interior decoration.
- Do not sit on laurels. Title and past scope depreciate faster when the production function changes. Relevance is a practice, not a plaque.
None of that is glamorous. All of it is insurance against becoming the middle layer the spreadsheet deletes first.
The org shape that follows
I expect organizations that work under AI-heavy delivery to look roughly like this:
- Executives at the top owning strategy — fewer, clearer bets.
- A lean middle layer owning specific business goals — not generic “people management” for its own sake.
- Highly productive ICs (and agent fleets) supported by real tooling — platforms, evals, review culture, operational ownership.
That is dramatically flatter than the multi-layer manager stacks many companies still run. Flatter is not kinder by default. It is less room for ambiguity about who owns the outcome. If your role only exists to pass information up and down, the system will route around you.
The question that matters
The question is not whether this shift is coming. Parts of it are already priced into how aggressive companies staff management. The question is whether you are preparing: technical depth, business literacy, and systems judgment — or optimizing for a job description that is evaporating.
What are you doing right now to stay relevant as the middle of the org chart thins?