You Don't Need to Be a Programmer to Use AI Like One

Before I get into anything, watch this video first. The title says it's about a note-taking app. Don't let that fool you. The note-taking is just the example he uses. What he's actually demonstrating is something much bigger. Watch it, then come back.

Watch this first

The note-taking use case is just one example. The underlying capability is what matters.

After two years of using ChatGPT daily, I switched to Claude late last year. Around the same time, I started experimenting with what that video demonstrates: multi-agent AI systems. The combination changed how I work, and I want to share what I've learned in plain terms, because most of the writing on this topic is either too technical or too hype-y.

The Two Limits Most People Hit

Most people use AI the same way: open a chat, type a question, read the answer. That works fine for simple things. But two problems come up quickly once you start relying on it for real work.

The first is that AI has no memory of who you are. Every new conversation starts from scratch. You're re-explaining your situation, your goals, your preferences every single time. The AI is capable, but it's operating blind.

The second is that AI responses have an effective length limit. If you ask one focused question, you get a thorough answer. If you ask ten questions at once, you get ten thin answers. The output gets spread across everything you asked, and none of it ends up being as useful as it could be.

Both of these problems have a practical fix.

A Team Instead of One Chatbot

The fix is to stop thinking of AI as one tool and start thinking of it as a team, where each member has a specific role and area of expertise.

I built two foundational agents to start. Alfred is my Director. Everything I want to accomplish goes through him first. He figures out what kind of task it is and who should handle it. Melody is my HR Director. When Alfred identifies a gap on the team, she creates a new specialist agent with the right background for the job.

Those two alone are enough to grow an entire team as you need it. Here's a real example of that happening.

Building a College Database

I needed to collect comprehensive data on hundreds of colleges for counselorsophie.com. I didn't Google it. I didn't hire a developer. I just described the project to Alfred.

Alfred routes the task. Stephano defines the role. Melody gets the brief.

Alfred recognized we didn't have a data specialist on the team yet, so he sent a brief to Stephano, my Senior Researcher, asking him to define exactly what kind of expert we needed. Stephano wrote a full candidate profile and forwarded it to Melody. A few minutes later, Melody had drafted a new team member named Iris, our Education Data Engineer.