The New Leverage: Why the People Who Direct AI Will Win

I've been using AI since the day ChatGPT launched — I was among the first users when it went public in late 2022, in the top 0.1% of early adopters. Not as a novelty, but because I immediately saw what it could become. Since then I've used it as a core part of how I work every day: running a consulting practice, managing government contracts, building a product business on the side. I've used AI to do things that would have taken me days — sometimes things I simply couldn't have done alone at all.

And somewhere in that experimentation, a pattern clicked into place that I think matters a lot — especially for people working in social services and nonprofits.

When personal computers arrived, a new kind of professional emerged. They could bend the machine to their will — building spreadsheets that replaced rooms full of accountants, automating processes that took teams weeks. They weren't necessarily smarter than their peers. They just had leverage. The computer multiplied whatever they put into it.

Then came the internet. The scarce resource shifted from computing power to information fluency. The people who could find, synthesize, and apply knowledge faster than everyone else had a new kind of edge.

We're in the middle of the next version of that same transition. And most people haven't noticed yet.

If you can make AI do what you want — clearly articulating problems, breaking complex goals into pieces, knowing when to push back on the output — you have the same leverage the power users had in 1995.

This isn't really about prompt engineering or being a tech person. It's about clear thinking made actionable. The best AI operators tend to be people who know what they actually want, can communicate it precisely, and can evaluate whether they got it. That's a deeply human skill. The interface just made it more visible.

Here's the analogy that clicked for me: think about how you give a project to an employee.

If you're vague — minimal direction, no context, no sense of what "good" looks like — you'll probably get vague work back. But if you give that employee time to ask clarifying questions, come back with what they've heard, push on the details until they really understand what you're after — you get a dramatically better result. Not because the employee got smarter, but because the collaboration surfaced what you actually needed.

AI works exactly the same way. The people getting the best output aren't just writing better prompts — they're having better conversations. They let AI ask questions. They refine. They push back. They treat the exchange as a process of building shared understanding, not a single request-and-response transaction.

For organizations in social services, this is actually good news. The sector is full of people with hard-won expertise, deep relationships, and genuine clarity about what good outcomes look like. That's exactly the foundation that makes someone effective at directing AI well.

Here's the more interesting question: if AI compresses execution the way computers compressed calculation, what becomes scarce next?

• Judgment When anyone can generate ten versions of anything in seconds, the person who can reliably recognize which one is actually good — and why — becomes disproportionately valuable. Not just aesthetic taste, but real domain knowledge applied to real decisions. Understanding your clients, your community, your regulatory environment. That doesn't get automated.
• Distribution & Trust AI can produce infinite output, but human attention is still finite. The organizations that have built genuine relationships with their communities, funders, and partners hold something that can't be prompted into existence. That trust accrues slowly and it's nearly impossible to fake.
• Accountability AI can draft the case note, the grant narrative, the program report. But someone still has to stand behind it. In high-stakes work — and most of what we do in this sector is high-stakes — the person with skin in the game still matters enormously.

What I find most interesting about this pattern is that each new layer doesn't replace the underlying skills — it elevates them. The best AI operators aren't people who stopped thinking critically. They're people who think clearly and now have a powerful tool to express that thinking at scale.

The transition isn't from human skill to machine skill. It's from doing the work to directing it.

The people who navigate that well are usually the ones who noticed the pattern while it was happening — not after someone else named it for them.

One practical place to start: your team. Getting staff past the resistance to AI tools is its own challenge, and it's mostly not about the technology. I covered the five fears that drive staff pushback and a week-by-week rollout plan in The People Wall: Why AI Rollouts Fail (and How to Fix It).