Hey Google, Meet Your Replacement

Every night, like clockwork, I ask my Google Home Mini to “play thunder sounds.” It’s part of the bedtime routine for my baby son — the low rumble of a storm helps him drift off.

I’ve added a few extras over time: “Set a timer for 90 minutes,” “What’s the weather tomorrow?” Sometimes it works. Sometimes it doesn’t.

What I’ve realised is this: despite its name, Google Assistant doesn’t really assist. It obeys simple commands and gets confused easily.

Try asking a follow-up question or anything with nuance, and it’s like speaking into the void. It’s less a digital companion and more like shouting instructions at a stubborn speaker.

Not just white noise

The contrast is stark when you use a tool like ChatGPT. Suddenly you’re having conversations that make sense. The AI remembers context, adjusts its tone, and even picks up when you’re being sarcastic.

It’s far from perfect, but it’s undeniably a leap forward from the “Hey Google” days.

That’s why OpenAI’s reported move into consumer hardware has my attention. The company has reportedly been working on a dedicated AI device, with help from none other than legendary Apple designer Jony Ive and a team of industrial engineers from LoveFrom, to build what might be the first true AI-first device.

It’s rumoured to be something ambient, always-on, and powered by GPT-4o, their latest model that can hold full voice conversations in real time, complete with tone and emotion.

Some speculate it could be a wearable, maybe an AI pin or headphones. But what I’m hoping for is something else: a smart speaker, but not the kind we already have. A proper assistant. One that feels like the sci-fi versions we grew up with.

You know the ones: the ever-calm voice of the starship computer in Star Trek, the softly spoken Samantha from Her, or even Alien’s “Mother,” who somehow always knew more than she let on.

A device you can speak to freely — not just to turn off the lights, but to ask questions, get advice, understand your world. Ideally one that’s helpful like JARVIS, not homicidal like HAL. (Although, to be fair, HAL did listen attentively.)

Rethinking AI in the home

That kind of assistant isn’t about flashy features. It’s about being useful in a human way — remembering what you’ve said, understanding how you speak, and responding with context.

Not a chatbot in a box, but a presence that lives in your home and actually helps. It may sound creepy, and we need a lot of safety precautions before this can work irl. ButI believe it can be done.

OpenAI might be the only company positioned to make it happen, at least for now. GPT-4o already responds to spoken questions with nuance. It can interrupt, joke, adjust its pace.

Imagine embedding that in a discreet, beautifully designed speaker on your desk or kitchen counter. You could ask how to save a cake batter while your hands are covered in flour. You could get a thoughtful response when your kid asks, “Why is Jupiter so big?” And it could all happen without a screen.

In an industry obsessed with disruption and doom, this vision feels refreshingly hopeful. Yes, there are serious concerns about AI: job losses, surveillance, synthetic media and the larger spectre of artificial general intelligence.

But not every application needs to lean dystopian. Sometimes it’s about making everyday things — like getting your baby to sleep — a little easier, a little calmer, a little more connected.

Of course, this is still speculation. OpenAI hasn’t confirmed exactly what it’s building. But the pieces are in place for something meaningful. And if it’s done right, this device could mark a turning point — from assistants that feel like novelties to ones that actually feel like assistance.

I’m not asking for a perfect AI. Just one that knows when nap time is. One that can help me think out loud. One that feels less like a product, and more like a presence — the kind we’ve imagined in fiction for decades.

We may not be on a starship. But maybe it’s time we got to talk to one.