Introduction
The manager's job has traditionally been described as directing people. In an AI-augmented organization, a more accurate description for many roles is directing a mix of human and AI-driven work toward a coherent outcome — a role best understood as orchestration, distinct from either pure execution or pure people management.
Why Orchestration Is a Distinct Skill
Orchestration requires capabilities that don't map cleanly onto either traditional management or traditional individual contribution:
- Workflow design judgment — knowing which parts of a process are genuinely better suited to AI tools and which require human judgment, and being willing to redesign the workflow around that distinction rather than simply adding AI to an unchanged process
- Maintained technical fluency — staying close enough to the actual work to evaluate AI output critically, rather than losing the hands-on expertise needed to spot when something is wrong
- Change communication — explaining clearly to a team which parts of their role are shifting and why, so the change feels like a deliberate design decision rather than an arbitrary or threatening disruption
The Risk of Losing Touch
A common failure mode for orchestrators is drifting too far from the actual work — delegating so much to both people and AI tools that the manager loses the contextual expertise needed to evaluate whether either is performing well. Orchestration requires staying close enough to catch problems, even while not personally executing most of the work.
Redesigning Workflows, Not Just Adding Tools
The most common mistake in adopting an orchestrator mindset is bolting AI tools onto an existing workflow without rethinking the workflow itself. A genuinely redesigned workflow asks, for each step: is this best done by a person, by an AI tool, or by some combination — and in what order? This is a fundamentally different exercise than simply introducing a new tool into an unchanged process.
What Good Orchestration Looks Like in Practice
- A team's workflow is explicitly mapped, with clear reasoning for which steps involve AI tools, which involve human judgment, and where the two intersect
- The manager can speak knowledgeably about both the AI tools in use and the human judgment being applied — not delegated fully to either
- Team members understand why the workflow is structured the way it is, not just what they're being asked to do differently