Launching Vox on Product Hunt

Tomorrow, April 7, 2026, I’m launching Vox on Product Hunt.

I’ve built a lot of things before this. Tools, experiments, side projects, and ideas that were exciting to make but easy to leave behind. Vox is not the first thing I’ve built, but it is one of the first things I’m trying to push properly into the world and stand behind in public.

The main idea behind it is simple: AI already feels normal, but AI products still do not.

My friends already talk to ChatGPT and Claude casually. That part is done. What still feels too technical is everything around the AI: accounts, API keys, model choices, provider choices, setup steps, permissions, and the general feeling that before you can use something useful, you first have to understand the plumbing.

That is the gap I keep coming back to.

I think the future of AI is not just smarter models. It is AI becoming more personal, more local, more private, and much easier to use. As models improve, inference gets cheaper, and personal machines get stronger, more useful AI should move closer to the user. Closer to your files, your context, your habits, and your actual computer.

That shift is interesting to me because it changes the product itself. AI stops feeling like a website you visit and starts feeling more like a layer of your computer. It can act, not just answer. It can remember, not just respond. And if it is built right, it can do that without constantly shipping your context somewhere else.

That is why Vox is voice-first, local-first, and open source.

I care about local because if software is touching your files, messages, notes, and screen, privacy cannot be an afterthought.

I care about simplicity because local AI still too often feels like a science project.

And I care about open source because personal AI should not ask for blind trust. If this stuff is going to become part of how people work, it should be possible to inspect it, audit it, extend it, and shape it.

In a loose way, I want Vox to slowly become something like the Linux of future personal AI. Not in the sense that it has to be for engineers only, and not in the sense that it should feel complicated. I mean open, user-owned, extensible, local, and shaped over time by the people who actually use it.

The challenge is making that future feel simple.

I do not want the product to win by exposing more knobs. I want it to win by hiding complexity better. Stronger machines should not create more setup. They should remove it.

So Vox is my attempt to build toward that version of computing: AI that lives closer to you, does real work on your machine, respects your privacy, and feels natural enough that normal people can use it without feeling technical.

That is why I’m launching it now. I do not want to keep convincing myself in private that the idea is good. I want real signal. I want to know if people actually feel this too: that the future of AI should be personal, private, open, and easy.

If you check out Vox, tell me what you really think. Tell me what feels useful, what feels confusing, and what still feels too technical. I’ve built a lot of things before. This is one I want to keep pushing.