AI can help a project team produce in an hour what once took days. It can analyze customer feedback, monitor dependencies, flag emerging risks, and generate options before the team has even met.
But faster work does not necessarily mean better-coordinated work.
At 8:07 on a Tuesday morning, a project lead opens her laptop to find an overnight alert: engineering may miss a critical milestone. Her AI tools have already analyzed the dependencies, identified the likely impact, and drafted several responses. By the time the team meets, the information is ready.
The disagreement is not about the data. Product is prioritizing retention. Engineering is protecting launch stability. Marketing is still working toward a date that may no longer be realistic. Each team is making reasonable decisions, but they are not working toward the same outcome.
As AI takes on more of the analysis and execution, the harder problem may be ensuring that people and agents are working toward compatible goals, making decisions with a shared understanding, and adjusting their work when priorities change. That is a coordination problem, and it is one organizations need to learn how to manage now.
So, what are we preparing people for?
Most conversations about AI and the workforce focus on individual capability: using tools, writing prompts, automating tasks. These skills matter, but a project is not a collection of tasks. It is a system of interdependent work, where one person’s output becomes another’s input and one change can affect the entire team.
Project risk has traditionally been managed in intervals: a risk register, a weekly status review, a follow-up on overdue tasks. Agents can now monitor continuously and escalate what needs human judgment. But AI can accelerate individual tasks, and detect individual risks, without resolving coordination. The biggest threat is less likely to be a missed task and more likely to be a team in which every contributor, human or agent, produces something reasonable based on a different understanding of what the project needs. That kind of risk spreads faster as work speeds up.
The next advantage may be coordination
Two organizations can have the same AI capabilities. In one, functions optimize their own outputs, risk is tracked by chasing late tasks, and conflicting priorities lead to another meeting. In the other, teams clarify shared goals, establish decision authority, make dependencies visible, and treat misalignment as a risk in its own right.
The difference is the team’s ability to direct that technology toward a common outcome. We don’t yet know exactly how AI-native teams will operate at scale, but decades of research on coordination, shared understanding, and adaptation give us a strong place to start.
Six capabilities to build now
- Define outcomes, not just tasks: specify what success looks like, what constraints matter, and how work will be evaluated, whether delegating to a colleague or an agent.
- Exercise judgment: test assumptions, weigh evidence, and know when human expertise is needed, especially when an agent flags or misses a risk.
- Build shared understanding: confirm what was decided and why. Distributing information is not the same as establishing agreement.
- Make dependencies visible: know who relies on your output and what changes when timing or assumptions shift.
- Clarify decision ownership: distinguish between input, recommendation, approval, and accountability, including who acts when a risk is escalated.
- Detect and correct divergence: notice when assumptions conflict and correct course before the consequences spread.
The takeaway
The organizations best prepared for what’s ahead won’t necessarily be those that deploy the most agents first. They’ll be the ones that manage risk as a team, not just a report. The future of work is about what people and AI can accomplish together, and we can build that capability now.
Sources
Salas, E., Linhardt, R., & Fernández Castillo, G. (2025). The science (and practice) of teamwork: A commentary on forty years of progress… Small Group Research, 56(3), 392–425. https://doi.org/10.1177/10464964241274119
Traylor, A. M., Tannenbaum, S., Thomas, E. J., & Salas, E. (2023). Teamwork in organizations: The best teams learn, adapt, and are resilient. In Principles of Organizational Behavior (3rd ed.), Chapter 18. https://doi.org/10.1002/9781394320769.ch18
Mayo, A. T., Woolley, A. W., John, L., March, C., Witchel, S., & Nowalk, A. (2025). Coordination in dynamic teams: Investigating a learning–productivity trade-off. Organization Science, 36(2), 967–992. https://doi.org/10.1287/orsc.2022.16729
