11/08/2026
A small moment that made AI agents feel very real to me.
Yesterday, I set up a robotics simulation environment on WSL with Ubuntu.
The usual process involves installing dependencies, configuring repositories, setting up ROS 2, installing Gazebo, fixing environment variables, dealing with errors, running tests, and debugging.
I expected to spend hours in the terminal, but I gave the task to an AI agent.
I explained what I wanted: “Set up my robotics simulation environment with ROS 2 + Gazebo.”
Then I watched it work.
It navigated the terminal, ran commands, installed packages, checked dependencies, detected issues, fixed them, configured the environment, and verified the installation.
I didn’t have to manually execute the setup step by step.
That’s when the concept of Computer-Use Agents clicked for me.
This wasn’t AI telling me how to install ROS. It was AI using my computer to do it.
From: “Here are the commands.” to: “Give me the task. I’ll execute it.”
For developers, engineers, and robotics people, this could be a massive shift.
The future isn’t just AI that writes code. It’s AI that can see, decide, execute, debug, and verify.
I think we’re only at the beginning. 🤖
