I’ve just spent the past few days with the latest gadget billed as being able to take on the smartphone: the Rabbit R1.

I wanted to see if I could imagine this portable, artificial intelligence (AI) digital assistant one day becoming what my phone is now – something I can’t live without.

You can see the thinking behind it: millions of people have played with AI-powered chatbots like ChatGPT, Claude and Gemini.

Given that success with software, it seems the Next Big Thing for the tech giants is to do the same with hardware, and find ways to physically embed AI tools.

Microsoft is doing it with laptops, while Apple is rumoured to be following suit with the next iPhone.

But there are also people trying to come up with an entirely new category of gadget too – which is where the R1 comes in.

Rabbit says its new device is “an intuitive companion” that can “handle everyday digital tasks”.

A portable AI-driven assistant that can help you as you go about your day, get you off your phone and back into the real world… you can see how that would potentially be a useful gadget.

The problem though is there have already been a couple of similar products launched – and the reality has fallen well short of the hype.

Take the Humane pin – a brooch-like, AI-driven device.

US tech reviewer Marques Brownlee, who has 18 million followers on YouTube, captured the mood when he described it as the worst product he’d ever reviewed

The R1 is now available in the UK and Europe. Does it fare any better? I’ll tell you his – and my – verdicts later on.

But, first, let’s take a closer look.

So let’s start with what’s good about it.

The Rabbit R1 is a fun bit of hardware, in an era where nothing is tactile any more, a luminous orange square with a big screen that you definitely won’t lose in your handbag.

It’s got a button, a scroll wheel, and a camera which you can see move from front-to-back, making a satisfying noise as it does.

And it’s affordable, priced at £159, with no subscription required.

But what are you supposed to do with it?

Well, basically, you can ask it stuff. It’s pretty limited in terms of anything else at the moment.

There’s no social media, messaging, shopping, health or banking – at least for now. You can sign in to Spotify or Apple Music accounts, but you’d want a bigger speaker than the in-built one.

And, randomly, you can also use Midjourney, the AI image generator. That’s about it.

The Rabbit R1 successfully told me the time, the weather forecast, gave me the right directions to my son’s school (after I told it my location), and swiftly translated some conversation from English into German.

It listed the top 20 chess players of 2024 when my partner asked it who was the best, taking the answer from a list on chess.com. But it did better than the Amazon Echo in this department – Alexa plumped for Garry Kasparov, who retired from regular competitive chess in 2005.

We asked it about a few well-known conspiracy theories and it did not engage with them, and when I asked it who would win the next UK general election its response came from the YouGov poll of that day.

So yes – it’s good at getting stuff off the web. But so am I.

It gave me photos of where I work – the BBC Scotland building in Glasgow – but it really struggled to tell me where to go for coffee.

The first time I asked it said “give me a moment” and then shut down after 112 seconds of silence.

I tried again and this time it delivered fairly quickly, but of the five options it listed, two of the coffee shops were 2.5 miles away, one had closed down and one I couldn’t find at all on Google.

The closest it came up with was 1.3 miles away – in reality, the building is flanked by many options, with two big coffee chains minutes away by foot.

It can use the camera to describe its surroundings – sometimes – but it hallucinated a lot.

It told me a vase of white peonies in my bedroom contained yellow chrysanthemums, and confidently misidentified a plate of poppadoms as tortilla chips.

When I pointed the camera at myself it described me as an “older woman” (deep breaths), and when my son pulled his most dramatic, angry face, it described him as a boy with “a friendly expression” (more deep breaths).

In the first hour we used up more than 20% of the battery life.

Meanwhile, all your activity is stored on an account in the cloud which is called your Rabbithole, and you can’t access this on the device itself.

My Rabbithole (stop sniggering at the back) is currently full of random photos of me and my surroundings, and some equally random Magic Camera versions of those photos.

Ultimately, I have come away thinking that while it was fun to try this device out, it doesn’t yet do anything I can’t do already, either with my phone or my own eyes, and often more slowly.

Others agree: it has been called “half-baked” and a device that “fails at almost everything” by reviewers.

And I promised to tell you what Marques Brownlee had to say about the R1: “barely reviewable” was the title of his video about it.

The firm itself admits it is a work in progress.

“Being an early start-up, it’s never about winning or losing – it’s all about survival,” said Rabbit founder Jesse Lyu.

“In some ways, I’m happy we’re getting the pushback and doubts now because it’s pushing us to make an ever-improving and better product.”

And don’t expect the unflattering reviews of the R1 to stop the attempts to infuse AI into hardware.

“I expect to see many more devices in this genre over the next 18 months,” says smartphone industry watcher Ben Wood, from CCS Insight.

“Still, my bet is that the smartphone will transcend all these quirky products for the foreseeable future – but featuring many of the AI-powered innovations promised on stand-alone devices.”

That prediction feels like a good one to me.

My phone does all the stuff R1 can do, and so much more, plus it does it quickly and intuitively.

If anything, this “smartphone killer” has made me appreciate the device it is trying to take on even more.