I’ve been checking out this “vibe coding” thing. Can’t tell you exactly what it is yet, other than AI helps you write code. There are a few players in the market. I’m not going to sit here and tell you I’m a developer. I’m not. I watched a couple of videos, threw some darts, and picked one.
Claude Code.
I’d heard of Claude for a while. OK, it’s just another AI. Insert ChatGPT here. Blah blah blah. Just another chatbot.
Then I read a reference about installing Claude Code locally, as a terminal application. I didn’t fully understand the difference between a terminal session and a web session at first. I read a little more. And a little more.
Here’s what clicked: if you install Claude in a terminal, it can see your network. A web-based AI is sandboxed, it’s abstracted from the operating system, which is abstracted from your network. A terminal-based AI sits on your machine and can interact with everything your machine can interact with.
You can ask Claude to do things.
I typed: “Hey Claude, ping 192.168.x.x.”
It responded.
Holy shit.
Three syllables. That was my entire reaction. I have an AI that I can speak to in natural language, and it can talk to things on my network.
I have never been so excited or terrified at the same time. My two thoughts, in order:
1. This can do everything. 2. Shit, this can do everything.
Right now, Claude’s guardrails seem to be on. I have to approve almost every local access request. There’s a lot of learning to do, a lot of guardrails to understand.
I know I sound like a broken record. This is another thing that changes everything. But I keep saying it because it keeps being true.
Here’s my honest assessment after about 10 days to 2 weeks: Claude Code is the smartest IT guy I know. Mostly because it can write Linux commands at approximately 1,000 words per minute.
That’s amazing. That’s also terrifying, for a guy who has to stop and research half of his Linux commands, having a partner who can type that fast makes me really, really fast. In my home lab. Under strict governance.
For the love of God, don’t bring this into your office until you understand what it does. Run it locally. In an environment that’s safe. Do your own research first.
I’m curious but skeptical about how this translates to a corporate environment. At home, my risk tolerance is higher, if my network breaks, nobody cares but me. At work, that’s a very different conversation.
This is a game changer. I’m not sure what game we’re playing anymore.