The Clarity Multiplier: AI In the Human Loop

I sat down with David O’Hara on the Improving Edge podcast to talk about the leadership skill of leading AI. We covered ground ranging from why I flip “human in the loop” on its head to what I believe is the real bottleneck in software teams.

Here’s a glimpse of what we talked about.

AI in the Human Loop

When people lean hard on the phrase “human in the loop,” I push back. It puts AI at the center and the human on the periphery, just watching. I think about it the other way: AI in the human loop.

Humans already have a loop. We have our meetings, our sprints, our ways of solving problems. The real question is where AI can plug into ours. It feeds into our loop when it helps, and it pulls back when we don’t need it. That framing keeps the human at the center, which is where problem-solving actually lives.

Clarity of Intent Is the Bottleneck

Much of the conversation came back to one idea: writing code faster has never been the bottleneck. Neither have the tools. You could always get better tools to write more code faster.

The bottleneck is clarity of intent. Human intent.

Most of us are bad at explaining our problems. We’re good at telling people what we want, but not why we need it, or even whether we need it. I lean on a simple framework: need, problem, solution. Figure out the need first. Then the problem. The solution comes last, and only if there still is one. Sometimes the need stops being a problem once you look at it from another angle, and there’s nothing left to solve.

That clarity is where AI earns its keep, if you let it help you think instead of just handing you output.

Stories, Not Requirements

We got into stories, which is always a favorite topic of mine. User stories are not requirements. They’re stories. They exist so people can say, once we’re done, here’s the story our stakeholders will go tell the world about why this thing is good.

And I’m careful about the word “user” in user story. A person is more than someone at a computer. If they’re working in a loud, distracting environment with customers yelling nearby, that matters. The beautiful UX design fails because it never accounted for the rest of the person’s world. So I think of it as the human story, not the user story.

Agents Are Just Recipes

People ask me about the agents I’ve built, and the most useful one has nothing fancy about it. It’s a markdown file. A recipe that captures how I write stories: how I identify the value, the persona, the context, the acceptance criteria. I built it by telling my AI to read years of my blog posts and write me that recipe.

An agent isn’t a magical thing. It’s a set of instructions that carries how you think, so you can feed it the same kind of input and get consistent output. I even rebuilt it for a colleague inside their Copilot, not by giving them my prompts, but by having the AI read the same blog posts and write its own.

Practicing Leadership With AI

My favorite part of all this is that AI gives people a place to practice leadership. You can’t put someone in charge of a team and say, go practice on these people. But you can practice on an AI. You learn to be clear about your vision and intent, to delegate, to ask it to plan its work, to watch where it stumbles, and to fill in the gaps. The Karate Kid stuff. Wax on, wax off. The reps matter.

The same skills you need to lead people are the skills that make AI useful. And if AI is just “here, do this because I told you so,” it fails. That’s a leadership problem, not a tool problem.

There’s more in the full episode, including the week’s biggest win for me: onboarding two new team members onto a project where every team interaction had been transcribed for months, and watching them get up to speed and contribute meaningful work to the business without me carrying the whole load. That’s the clarity multiplier in action.

Watch the full conversation with David on the Improving Edge:

Claudio Lassala's Blog

Leave a ReplyCancel reply