The web was meant to free us. Instead, we rent our memories back from tech bros.

I’ve been thinking a lot about the AI assistant I want to build.

Not a chatbot. Not a feature. A real assistant—a second brain, a memory prosthetic, a true extension of my cognition.

Something that: - Sees what I see, hears what I hear (when I say so) - Remembers what matters—chronologically - Helps me reflect, query, plan, understand - Tracks my finances, supports my work, runs my tools - Lives on my machines. Not in someone else’s cloud. - Belongs to me—and only me.

And you know what? The tech is there.

Local LLMs. Open-source models. Vector DBs. Voice. OCR. Containers. MQTT. ESP32s. We can build this.

But everything in our current ecosystem is designed to prevent us from doing so.

• iOS won’t let you access the mic during a call
• Browsers won’t let you remember your history unless you hand it to Google
• Files and photos live in silos
• Apps guard their little fiefdoms like medieval barons

You can’t build a memory prosthetic on a foundation of walled gardens.

This is not a technical problem. It’s a business model problem. A control problem. And it’s making technology dumber, not smarter.

We’ve ended up with “AI” that can write your essay but forgets you the second the tab closes.

⸻

Here’s what I’m dreaming of instead: - No apps. - No ads. - No lock-in. - Just a device—a mic, a camera, a screen. - A minimal OS. No app store. Just APIs. - An LLM that draws its own interface when needed. - You talk, you show, you type—it responds, recalls, reflects. - All local. All yours.

Forget “there’s an app for that.”

APIs will replace apps. Context will replace clicks. Models will draw their own UI.

That’s where we’re headed. The only question is: Will anyone build the hardware to match this shift?

Is there a single company out there brave enough to ship: - A clean-slate device - With an open LLM + API layer - That puts the user—not the investor—at the centre - And says: this is yours, do what you want

Is it Rabbit? Doubtful. Is it Apple? Not with their current DNA. Is it… us?

⸻

I’m building something myself now. A Docker-based, self-hosted assistant with voice, memory, vision, tool use, cost tracking, and multi-tenant architecture. It’s not ready for the world, but it’s real. And it’s mine.

Because honestly, I’d rather build that future than wait for a trillion-dollar company to sell me a lobotomised version of it for $29.99/month.

Who else feels this? Who else wants to reclaim their cognition?

Let’s talk.