Day 1: Starting My AI Mastery Journey

I’ve been curious about AI for a while, but curiosity without action stays curiosity. Today, that changed.

The Setup

I work across two machines: a MacBook Pro (M1 Pro) for daily development and an ASUS ROG Flow Z13 with 128GB RAM for heavy AI work. The Z13’s specs—128GB RAM, 16 cores—let me run large local AI models without cloud dependencies. That matters to me.

First decision: document everything from the start.

Not as tutorials or guides—I’m not teaching anyone yet. This is my lab notebook. If someone else finds value in it later, great. But the primary purpose is to think clearly and track my progress.

The Context Problem

Working with AI assistants, I noticed something: they forget. Every new session starts fresh unless you feed them context manually. That’s inefficient.

I created two files:

  • CLAUDE.md - Technical context for AI sessions
  • JOURNEY.md - Personal narrative for learning

The AI reads CLAUDE.md and immediately understands where we left off. No re-explaining. No “as I mentioned earlier…” Just continuity.

Simple idea. Surprisingly effective.

First Systems Built

1. Git-Based Timeline

I initialized a git repository for my journey. Every significant change gets committed. My progress has a timeline now—searchable, reversible, portable.

Connected it to GitHub (private for now). If my machine dies, my journey survives.

2. Local Speech Recognition

I wanted offline voice-to-text. Privacy matters, and I didn’t want to depend on cloud APIs.

Installed Whisper (OpenAI’s speech recognition model) locally. Built a push-to-talk system with Python:

  • Hold Alt key → speak
  • Release → auto-transcribe and paste

Works across any application. No internet needed.

3. Cross-Machine Context Sync

I work on a MacBook Pro and this Z13. Switching machines meant losing context.

Built a solution using MCP (Model Context Protocol) servers. Now I can set context on one machine and retrieve it on another—instantly, via cloud sync.

The infrastructure is there. The workflow is seamless.

What I Learned

Systems thinking > individual tools.

Anyone can install software. The real value is in how you connect things together. Documentation systems, version control, cross-machine sync—these aren’t fancy, but they compound.

Document as you go, not after.

Trying to remember what I did three days ago? Hard. Having a git log and journal entries? Easy.

AI amplifies your workflow—if you build the right scaffolding.

Claude Code is powerful, but only when it has context. Feed it structure, and it becomes a thinking partner instead of just a chatbot.

What’s Next

I’m exploring Model Context Protocol more deeply—specifically how to give AI access to:

  • My filesystem (securely)
  • My git repositories
  • My GitHub projects
  • Cross-machine shared state

If I can build an environment where the AI “sees” what I see, collaboration becomes more natural.

That’s the experiment for the next session.

This is part of my journey from curiosity to AI mastery. I’m documenting everything I build and learn along the way.