Your team is changing
For most of organizational history, a team has meant a group of people working toward a shared goal, but that assumption is beginning to change.
AI agents can now research a problem, monitor a workflow, analyze information, create an output, or take action across systems. As agents become more capable and autonomous, many teams will increasingly include both humans and AI agents.
This raises a fundamental question: What makes an agent a team member?
We don't necessarily need to redefine "team" from scratch. Decades of research define teams less by who their members are than by how they work together: they have shared goals, differentiated roles, interdependent work, and coordination. This same logic can apply to an AI agent.
An agent starts to look like a team member when it has:
- A role: a defined responsibility within the team's work.
- Interdependent work: what it does affects what other members do.
- Information others rely on: it produces or maintains information that matters to the team.
- Impact on outcomes: its actions can change the team's results.
- Some decision authority: it can recommend or take action within defined boundaries.
Consider an agent responsible for monitoring project risk. It analyzes project data, identifies emerging risks, prepares a daily brief, and escalates issues requiring human judgment. In this case, the agent has a role within the workflow, and other members depend on its work which is no different from a human teammate.
So, how teams work also needs to change
Once an agent becomes a participant in the work, the questions become familiar team-design questions: What is it responsible for? What does it need to know? What decisions can it make? When should it involve a human? How does the team know what it has done?
Research on human-agent teams reinforces many of the principles we already know from human teams: coordination, communication, shared mental models, role clarity, situation awareness, and calibrated trust all matter.
We have decades of research on how humans work together and a growing body of research on humans working with autonomous systems, giving us important foundations to build from. But the enterprise is now asking us to apply those lessons in a very different environment.
So what do we actually need to change?
- How should accountability work? The presence of an agent doesn't eliminate human accountability. Someone still needs to own the outcome, but that ownership may shift from doing the work to defining the goal, setting boundaries, reviewing performance, and intervening when necessary.
- How much autonomy should an agent have? The right question is what the agent is capable of doing reliably and what level of risk the team is willing to accept. An agent might observe in one situation, recommend in another, and act independently in a third. Research on human-agent teams increasingly points toward deliberately designing the allocation of authority rather than simply maximizing autonomy.
- How does a team know whether to trust an agent? Teams need calibrated trust based on evidence of the agent's capabilities, limitations, and performance. Humans also need a useful mental model of what the agent is doing and why.
- How does the team learn from an agent's decisions and mistakes? This may be one of the biggest shifts. An agent's work can't simply disappear into a workflow. Teams need visibility into what the agent did, what information it used, where it was uncertain, and whether its decisions produced the intended outcome. That creates a feedback loop for improving both the agent and the way the team works with it.
The takeaway
AI-native teams don’t just use AI in their daily work; they are teams in which work, information, and increasingly decision-making are distributed across humans and machines.
We don't have all the answers yet, but we know enough to see that the challenge is learning how to design, coordinate, and lead teams that include non-human members. That begins by building on what we already know about effective human teams while developing new approaches for working with autonomous agents.
The future of work is about learning how humans and AI can work as a team.
