Every Lesson I Give Claude Code Compounds

Every Little Node Is Something I've Worked On

The video opens on a visual representation of the second brain my Claude Code assistant uses to keep all of its information organized and connected. Every little node is something I've studied, a project I've worked on, or a prompt I've given it to help with my business operations work. The connections between them are like the connections in your own brain, and that's how I programmed it.

Anytime I'm working on a project, working with a client or doing something for myself, it remembers everything I do. Every detail, every process, even every mistake it makes.

It takes a while to populate, which is why it moves around in the video. Once it settles you get a static brain, and you can move the nodes around if you want to play with it. Connected to Claude Code, it gives you a picture of how your memory works and how the topics relate to each other.

Every New Session Was A New Person

This is probably the fifth iteration of the system. In the first one I wasn't even using Obsidian. I just told Claude Code I was running into a problem: I had to reprompt it every time I wanted to run a similar project, which was time consuming.

Each time you open a new session in Claude on your desktop, it's as if you're talking to a new person with very little context about you or what you're working on. You have to remind it, and you have to be good at prompting to get good outputs.

This sort of bypasses that. It wakes up with rich context about who I am, what I do and the problems I've been solving, and it learns each session. It keeps a daily log of what we've been working on and the things it needed to remember.

I've also built a lessons prompt into my Claude Code. Anytime it makes a mistake, I just say, "Hey, remember this lesson. You did this wrong." And it remembers.

The Second Time Is A Lot Easier

Your memory is very limited compared to what this technology can do.

Say you're working with a client on something new, like helping them optimize their CRM, or you're creating a new workflow for a video project, and the steps are new to you. If you program Claude Code to do this, and you certainly can, it remembers every step along the way so you don't have to. The next time you work on a similar project, the steps are already laid out for you and for your assistant.

Think about the first time you do anything. It usually takes much longer because it's unfamiliar, and there's a learning curve. If you're like me, the second time is a lot easier. It's the same thing with this, only better, because human memory is selective and emotional, and this doesn't have any of that.

Every project and every lesson you give your assistant keeps compounding, and the work gets faster and more efficient over time with you having to do less.

Claude Code In A Terminal

I'm working in a terminal. It works on things for me while I do what I need to do, and I can open up to four terminals inside one terminal and work on things at the same time.

Maybe eight months or a year ago I had no idea what this was, and now I love it. It's the top level of AI workflow you can incorporate into your business and into your life, and that's what I help people do.

This is the superpower of our age.