Here’s a quick quiz.
I’m going to give you a short list of capabilities. All I want you to do is tell me what they’re essential for.
• Perspective-taking
• Providing context
• Testing for understanding
• Asking follow-up questions
• Giving and receiving feedback
• Building shared history
• Relationship building
If you said “collaborating with other people,” you’re absolutely right. These are the cornerstones of social intelligence—the capabilities that separate productive teams from dysfunctional ones, effective leaders from ineffective ones, and successful negotiations from failed ones. I’ve spent more than two decades researching and teaching these skills to executives, founders, and teams across five continents. Perspective-taking—the ability to step outside your own head and imagine how someone else sees the world—is among the most robust predictors of collaborative success in organizational research. When I teach this, I emphasize that your collaborators don’t have access to your intentions. They can’t read your mind. You have to bridge that gap deliberately.
So far, none of this is new.
Plot Twist
Every one of those skills is now just as essential for collaborating with your AI agents.
Think about it. How many times have you muttered at Google Search, or Siri, or Alexa: That’s not what I meant. That’s not what I wanted. That’s not what I was trying to do. We’ve all been there. And most of us blame the technology. But here’s what I’ve come to realize: the real problem is a failure of social intelligence—directed at a non-human collaborator.
Consider the humble Google search. When search engines first emerged, the people who excelled at finding information weren’t necessarily the most technically skilled. They were the ones who could take the perspective of other humans—imagining what keywords someone else likely used when creating or describing the content they were looking for. It wasn’t a technical skill. It was perspective-taking, plain and simple, applied to an information system. The Dewey Decimal System didn’t ask this of you. Google did.
Working with AI agents requires the same cognitive move, but deeper. You have to break through what psychologists call naïve realism—the default assumption that others see the world the way you do—and recognize that your AI agent cannot fully access your intentions, your motivations, or the context behind your requests. It doesn’t know what meeting you just walked out of, what’s weighing on you this week, or why this particular task matters to you right now. In other words, your AI agent is a lot like a brilliant new colleague on their first day. Immensely capable. Lacking context.
The Same Playbook, New Collaborator
Once you internalize this, the playbook writes itself—because it’s the same playbook you already use with the humans around you.
Provide context. Don’t just tell your AI agent what you want. Tell it why you want it, who it’s for, and what good looks like. Brief it the way you’d brief a talented new team member—not bark orders at a vending machine.
Test for understanding. After giving your AI agent a complex request, ask it to summarize what it thinks you’re asking for before it starts working. This is the AI equivalent of “can you play that back to me?”—one of the most underused and most powerful moves in human collaboration, and just as transformative here.
Ask follow-up questions—and invite them. The best collaborations are dialogues, not monologues. You can literally ask your AI agent what additional information would help it do a better job. Most people never think to do this. It’s a small move that changes everything.
Give feedback. When the output isn’t right, don’t just start over from scratch. Tell the agent what worked, what didn’t, and why—the same skill that makes you effective in a design critique or a coaching conversation.
Build shared history. Just as the best human collaborations improve over time through shared understanding, you can build that same kind of working relationship with your AI agents. Document what works. Save your best prompts. Create instructions that capture your preferences and working style. And here’s a genuine superpower that has no human equivalent: you can copy and paste this shared context into an entirely new conversation—instantly onboarding a new collaborator with all the history intact.
Get outsider feedback. There’s a move so counterintuitive it deserves its own moment: you can ask one AI agent how best to communicate with another. You can use an AI collaborator to help you articulate your own thinking more clearly—refine your request, sharpen your framing—before handing the task to a different agent optimized for that work. It’s like having a communications coach on call, 24/7, who never gets tired of helping you say what you actually mean.
Human Mode × AI
There’s a deep irony here worth sitting with. As intelligent machines, large language models, AI agents threaten to marginalize human capabilities, the path to thriving doesn’t require simple technical acumen. It requires you to double down on what makes you most human.
The greatest gains don’t come from AI capability alone, and they don’t come from human capability alone. They come from the interaction between them—what I’d call the Human–AI Multiplier. And the quality of that interaction depends on distinctly human skills: the ability to see the world through another’s eyes—even when those eyes are artificial—and to communicate with the clarity, empathy, and intentionality that great collaboration has always demanded.
As AI handles more of the routine cognitive work, the remaining contributions shift precisely toward judgment, creativity, and relationship management. The very capabilities the old model told us to leave at the door.
Social intelligence has always been the capability that separated good collaborators from great ones. That hasn’t changed. What’s changed is that your list of potential collaborators just got a lot longer—and some of them are made of silicon. The question isn’t whether you have the skills to work with AI agents. You already do. The question is whether you’ll recognize that the deeply human capabilities you’ve been developing your entire career are exactly what this moment demands.
This is human mode meeting the machine age. And it turns out, human mode is the edge.