Hey HN! I’m Andrew, and I’ve been building Cortex (https://cortexdesktop.com/) for the last few weeks. Cortex is my attempt to reimagine how we interact with computers - starting with a simple interface, but aiming for something deeper: an AI layer over your OS that understands what you’re doing, remembers it, and helps you dynamically.

Right now, Cortex is just a floating overlay you can summon anywhere on your Mac. It works like an inline ChatGPT, but it lets you instantly send that context into the chat by highlighting text - so you don’t have to explain what you’re doing. Here is a demo: https://www.youtube.com/watch?v=fZEbpysqH2w&feature=youtu.be

That’s all it does today but the reason I’m building it is because I think that the current way we use computers is kind of antiquated - the OS is relatively dumb and memoryless. I feel like it would be so much better of an experience to work with a computer that is contextually and semantically smart.

This started as a tool I was hacking together for myself while context-switching between projects, tabs, and apps. I was using ChatGPT a lot, but every time I wanted to ask something, I had to re-explain what I was doing. “I’m building this project with (these) specs, and (this) stack, help me plan etc etc”

I realized that firstly, it was super tedious to constantly go back to GPT to query, and secondly that it was annoying that I constantly had to feed it context to make it work the way I wanted. Granted, Chat GPT does have a certain amount of memory nowadays but I mean something far more granular - file diffing, text changes, differences in entire workflows, etc.

I feel that the OS should have memory. It should be able to understand semantically, not just react to commands. Eventually, I want Cortex to become the operating system’s memory. A system that watches what you do across apps (files, text, window titles, tabs), builds a real-time semantic timeline of your work, and lets you ask questions like “What was I working on Thursday before lunch?” Or say “Open the Figma file I was editing after that Zoom call”.

I am technically getting ahead of myself, as right now I’m just trying to test the one use case of having access to AI anywhere in the OS. This is definitely an early-stage product. It’s glitchy in spots. But if you like trying new interfaces and imagining what comes next - I’d love your feedback.

You can use it free for 7 days — no signup needed. I’d love to hear your feedback, ideas, critiques, or even thoughts on how you’d want your computer to "remember."

Thanks!

-Andrew