Managers, get ready: You\u2019re about to lead a new team. But your new team isn\u2019t human \u2013 it\u2019s a team of AIs. And you\u2019ll be accountable for their work.<\/p>\n
The era when \u201cteam leadership\u201d meant only humans is over. The savviest workers are now using multiple AI models. Managers now need to decide which model is assigned to which task, delegate tasks effectively, and review the output from each. That sounds a lot like leading a team.<\/p>\n
AI use is exploding. Microsoft\u2019s 2025 Work Trend Index<\/a> reports that 36% of leaders expect managing AI systems to be part of their scope within five years. The World Economic Forum hears CEOs predicting future leaders will be measured partly by how many \u201cdigital workers\u201d they can effectively manage.<\/p>\n Luckily, not all the skills for leading a team of AI agents are unfamiliar. While there\u2019s some learning to do, you can use some basic principles of teamwork and leadership to help guide your new team of AIs \u2013 with a few tweaks.<\/p>\n What stays the same \u2013 and what changes?<\/p>\n Using AI well is no longer about picking the right app. It\u2019s about orchestrating a set of agents, each with unique strengths and weaknesses. Like any team, they can underperform when misaligned\u2014or create unity when structured well.<\/p>\n Research from Harvard\u2019s David Deming and colleagues<\/a> shows a strong link between skill at coordinating AI agents and skill at leading human teams. The same leadership moves matter: asking clarifying questions, setting clear expectations, and learning through trial and error. Microsoft <\/a>calls this the rise of the \u201cagent boss,\u201d urging leaders to onboard AIs like new hires\u2014compose your \u201cteams\u201d carefully, set clear expectations, monitor early outputs, and establish trust-but-verify routines.<\/p>\n That\u2019s why the basics of leading your AI team are the same as leading a human team: establish clear goals, define roles, and provide performance feedback\u2014themes I talk about in my book The Collective Edge: Unlocking the Secret Power of Groups<\/a>.<\/p>\n However, leading a team of AIs is not the same as leading people. AI agents don\u2019t negotiate roles, learn from each other, or self-organize unless explicitly programmed to do so. There\u2019s no shared leadership, no emergent norms, no psychological safety\u2014but also no fear or status dynamics holding them back.<\/p>\n Years ago, Harvard professor J. Richard Hackman argued<\/a> that team leaders should be less like orchestral conductors and more like leaders of small jazz ensembles, with lots of room for creativity, shifting leadership, and improvisation. But, with teams of AIs, leaders are more like orchestral conductors. All coordination is centralized in the human.<\/p>\n AI also demands more technical fluency than human leadership. As Gabi Beyo notes<\/a>, you can\u2019t rely on charisma to get an AI back on track; you need to understand prompt construction, error patterns, and model capabilities.<\/p>\n From our research, here are the five leadership skills for you\u2019ll need to manage teams of AIs:<\/p>\n Team Composition and Role Clarity: Stay up to date on the latest models and their strengths and weaknesses. Match model strengths to task requirements.<\/p>\n Delegation and Task Design: Delegation is a core leadership competency, whether you are working with humans or LLMs. You define the task (prompting), choose the right \u201cteam member\u201d (model\/tool), and provide context with the clarity you\u2019d give a smart junior hire.<\/p>\n Performance Monitoring and Feedback: Build iterative refinement into the workflow. A common practice is to have one AI give feedback on the work of another. AI, like a new team member, often needs multiple tries.<\/p>\n Trust Calibration: You can\u2019t trust today\u2019s AI models to work unsupervised on unfamiliar tasks. Trust, but verify \u2013 especially for new tasks.<\/p>\n Ethical Oversight: As Ayumi Moore Aoki emphasizes<\/a>, always verify that outputs align with values.<\/p>\n The leadership memo you didn\u2019t know you needed<\/p>\n