I am addicted to technology since I can remember.

But my parents had a lot of care to protect me against vices and harmful things, in early age, my contact with technology was short and measured. I had my first own computer with around 13 years old, and my first smartphone only around 16. I feel highly grateful for this care.

Unfortunately my issue was games. The whole day in the computer, exiting only because of the hungry, sleepiness or headache… And doing nothing useful.

I was aware of that idiotic behavior but it was hard to solve.

Then around my 15 years old, I heard the pressure, advises and incentives from my father, and I started to do something useful, I initiated my journey on Linux and programming through a expensive yet shallow course about pentesting.
It made me realize that I like the area, but without programming, I would never get rid of the label of "script kiddie".
So after some a minimal comfort at Kali Linux by following the course, I saw a opportunity to get rid of my vice.

I abandoned Kali Linux from my VMWare and installed a pure Debian with dual boot on my notebook, previously with Windows 8.1.

After it, this became my new game.

Some time later, more than a year I removed Windows, and sticked on Debian for a long time. Got relieved, and games was no longer a problem!

Warping time back to now.

One of my values is to avoid distractions. Distractions are hard to handle, pleasing to the moment but packed with a high potential for damage.

What I see from LLMs is convenience.

Convenience a complex thing. To be convenient it generally not requires much effort, and with LLMs, we're now able to outsource even the intellectual efforts, … well, I think it might be a problem.

Reality is often a mix of the both sides, despite the influence and biases, you can still hold control of LLM output. But the second part is worse: privacy.

LLM is only useful with a deep context, and considering this technology is extremely resource expensive and you use a third-party provider, it becomes a intentional spyware in your computer.

I am writing this because I entered too deep on it, started to get lazy and avoid reading the codebase in order to understand. It started to build me the habit to avoid reasoning. I had enough, this is not the right way to use a tool.

It's hard to get rid of it. I am telling this words after using it a lot to deliver a full refactor on a legacy React web app while learning to use the framework in a little more than 3 months at work. LLM indeed helped, but in the end, the satisfaction of finishing wasn't even as expected.

So now, until I get the enough computation power to run a conversational LLM locally, I will use the same strategy I use for Google, YouTube, social media and any kind of threat for my time and privacy: Sacrifice the convenience.

Considering the few situations where LLM really help and don't messes the codebase (like code generation instead editing), I still rather spend some extra minutes reading documentation and learning normally as I always did: Facing frustrations, getting stuck and then get relieved and proud when overcome the problem. This builds a solid experience and makes me more efficient to solve future problems.

This is the process I love, and I rather should depend on no online service, so I can continue working normally without internet.

And for sure, this is a prerequisite for my airgapped environment.

For now, under my current circumstances and mindset, I rather get overwhelmed with problems, search at Kiwix snapshots for solutions and run ripgrep on cloned repos for keywords that brings solutions for my problems.

This is what worked for now, and this doesn't requires me to have a computer which is more expensive than all I spend in two years living in this world. But some day I'll have a offline supercomputer running my own conversational LLM :)

Tough decisions are being done, so here's the complete timeline tracking my usage of LLMs in my workflow and personal life.

timeline:
  - name: Playing with Google Bard and Nim
    date: 2023-07-18T00:33:57+00:00
  - name: Playing with ChatGPT and Nim
    date: 2023-08-21T19:43:20+00:00
  - name: Setupped Ollama local server with Llama 3.1 7B
    date: 2024-07-26T10:03:38-03:00
  - name: Installed Ellama at old Doom Emacs config repo for Ollama
    date: 2024-07-27T09:52:59-03:00
  - name: Installed Copilot Mode at old Doom Emacs config repo
    date: 2024-12-20T16:11:17-03:00
  - name: Configured Aider at old dotfiles repo
    date: 2025-01-30T10:07:17-03:00
  - name: Configured opencode
    date: 2025-08-05T10:24:02-03:00
  - name: Tried Claude Code
    date: 2025-08-14T09:45:15-03:00
  - name: Removed Copilot Mode completion from Emacs
    date: 2025-09-28T08:57:46-03:00
  - name: Removed Aider in old dotfiles repo
    date: 2025-10-17T20:51:11-03:00
  - name: Removed opencode from dotfiles
    date: 2025-11-13T16:30:08-03:00