The next era of work isn't just about employees using AI—it's about employees working alongside agentic AI. And the shape of that collaboration is starting to come into focus.
WSJ reporter Isabelle Bousquette sat down with Amazon, Coinbase and EY consulting to hear how some of the bigger companies are integrating agentic AI at scale.
It will look like a complete “restructuring of engineering teams into smaller, more nimble cross-functional ‘pods,’ made up of humans and AI agents” she writes based on her interviews and research. There won’t always be one agent for every human but there will be a few like co-workers.
Why it matters for CHROs: This isn't just a tweak to the org chart—it's a preview of how roles, headcount, and talent strategy will shift across the enterprise.
What makes a good pod:
-
A working group of 1-8 humans, plus AI agents
-
A cross-functional team of engineers, designers, applied scientists
-
AI agents do more of the coding and testing
-
Humans keep the project on track and refine outputs
Two caveats: Not every large company can follow this structure, at least not yet, because it requires “AI-native talent” who can control “fleets of agents to drive outsized impact.”
If a company doesn’t have an AI-native pipeline of talent they will likely need to create bridge roles meaning people who sit between the broader workforce and the agents, translating business needs into agent workflows, overseeing agent output, and coaching employees on how to use them. Think of it as a conduit layer: it buys time to upskill the rest of the organization while still reaping productivity gains.
Past parallels:
-
This pod structure is not entirely new. It’s “just the next evolution of the scrum team,” according to Dan Diasio, Ernst & Young’s global consulting AI leader.
-
It’s also reminiscent of Amazon’s “two-pizza team” rule although now, a 16-person team might split into two pods of eight,” notes Amazon VP Deepak Singh.
CHRO questions to ask:
-
What would small AI-augmented teams look like and in what areas of the organization?
-
What criteria and skills are needed for “AI-native talent” and planning for that in hiring?
-
How do you redesign roles when an engineer, designer, and project manager collapse into one?
-
How do you assess performance and pay people who manage agents instead of people?