Amazon reboots Alexa – the voice assistant that now knows you better than yourself

First Amazon assumed creative control of James Bond, now it's racing ahead in the battle to control people’s homes, unveiling its long-awaited overhaul of its Alexa voice assistant. Here is what’s new.

Amazon has overhauled its Alexa voice assistant to make it smarter and more conversational to reignite its struggling devices division and remove the friction from speaking to a computer to perform everyday tasks.

The $US2.28 trillion ($3.6 trillion) company, which last week assumed creative control of the James Bond film franchise, says it has “completely redesigned” the 10-year-old assistant, rebranding it Alexa+.

It can order groceries for households, remember personal details such as family members’ dietary preferences and favourite movies, become a “virtual security guard” and better co-ordinate smart home devices among its myriad features.

Amazon chief executive Andy Jassy said Alexa+ – which will cost $US19.99 a month and be free for Prime members – had made the voice assistant “meaningfully better” as the company vies for dominance in the $US1.35 trillion connected home market.

Mr Jassy said what made Alexa+ different was it tapped into generative – or conversational – artificial intelligence, allowing people to speak to the assistant in plain English and more naturally to perform every day and household tasks.

“Our team has been working on this for several months. I have had the chance to use it for the last several weeks. It’s really remarkable in my opinion,” Mr Jassy said in front of a select group of journalists in New York.

“We don’t use technology because we think it’s cool or interesting. We use it to solve real customer problems.”

Real-life Star Trek computer

Amazon launched Alexa in late 2014, taking inspiration from the Star Trek computer – an omnipresent present and proactive artificial intelligence-powered machine that people could control with their voice.

“However, up until a couple of years ago, it was pretty difficult to invent with AI, and that changed with the arrival of foundational models in generative AI and made this technology more accessible,” Mr Jassy said.

Amazon is undercutting its competitors by up to 75 per cent as it seeks to take the lead in the artificial intelligence race and overcome one of the biggest hurdles in launching the technology at scale: cost.

“We have a deep partnership with Nvidia. We have for a long time, we will for as far as I can forecast out in the future,” Mr Jassy said.

Amazon CEO Andy Jassy says Alexa+ is 'meaningfully better' than the original version.

“But our internal builders and our external customers want better price performance, which is why we have built our own custom AI silicon. As you get these generative AI applications to scale. They have to have the latency and have the right customer experience, and they have to have low cost.”

Undercutting rivals

Late last year, it released six new foundation models after the tech behemoth was perceived as falling behind rivals Microsoft and Google in the AI boom. Three of Amazon’s new models were “at least 75 per cent less expensive than the best-performing models in their respective intelligence class”.

Alexa+ will tap into these models as well as others available on Amazon Bedrock – a platform, which includes models from companies including Anthropic, Meta, Mistral AI and more recently China’s DeepSeek. 

Alexa’s overhaul is the first big product launch from Amazon’s devices and services business boss Panos Panay, who joined the company in late 2023. Mr Panay was previously at Microsoft for more than 19 years, where he launched Windows 11 and led the creation of the Surface line of computers.

He has been tasked with reigniting Amazon’s devices business, which between 2017 and 2021 racked up more than $US25bn in losses. This is despite more than 600 million Alexa-capable devices being in use globally.

Amazon’s devices and services business boss Panos Panay launches Alexa+ in New York.

What’s new?

Mr Panay likened Alexa+ to a symphony orchestra. “If you’ve ever been to a symphony, when you first get there, the orchestra starts warming up. Each instrument is pretty much tuning to perfection. It’s lovely, but it’s also an isolated moment and without the greater context, the notes have very little cohesion,” Mr Panay said.

“Then there is this moment when the conductor raises their baton, in a single gesture, all the pieces are then brought together, transforming disconnected sounds into a masterpiece. The new Alexa knows almost every instrument in your mind, your schedule, your smart home, your preferences, the device you’re using … and brings them together into what is an incredible centre.”

“She’ll find all your smart devices to learn the rhythm of your life and proactively take action with you.”

Mr Panay said this is what differentiates Alexa+ from the “explosion” in AI tools, apps and features.

“Sometimes they can get you started on a pretty good task. Most of the time, they won’t help you complete that task. In general, this is because they are isolated from other parts of your life. They don’t connect to apps and services you use every single day.

“Then there is another group where AI can actually be intimidating. It’s often hard to know where to start and hard to know where to go.”

More conversational

But he said with Alexa+ all people needed to do was speak naturally. To showcase this, he asked Alexa what was the song that Bradley Cooper sang in a duet. Alexa immediately answered that it was Shallow with Lady Gaga from the movie a Star is Born. Mr Panay asked Alexa to play that track on the TV and then the scene from the movie.

He also asked Alexa what had been happening around his house in Seattle, if he had any parcels delivered or anyone had walked his dog. The assistant gave a summary based on footage recorded on Ring doorbell and cameras, which Amazon owns, and took Mr Panay to exact clips, becoming a “virtual security guard”.

Amazon’s Alexa voice assistant was envisaged to work like the computer in Star Trek.

“It’s just meant to be natural. You don’t have to think there’s no more Alexa speak,” he said.

“She’s your virtual security guard, your personal shopper, she’s your music, sports experts, and so much more.”

Mr Panay said the more people use Alexa, the smarter the assistant gets. But it’s not just gleaning information from the internet and household connected devices.

People can upload documents, from handwritten recipes from their grandmother to “dense” legal documents such as council bylaws, and ask Alexa questions about them such as can I install solar panels on my roof or how much oil did grandma put in her zucchini bread.

Complete re-architecture

Daniel Rausch, Amazon’s vice president of Alexa and Fire TV, said Alexa+ had undergone a “complete re-architecture”.

“It is not as easy as taking an LLM (large language model) and jacking it into the original Alexa. That is not what’s being done here. It’s a complete re-architecture that has never been done at this scale. And it’s the largest integration of services, agentic capabilities and LLMs that we know of anywhere.”

Agentic AI is a type of artificial intelligence that can make decisions, take actions, and learn on its own.

“Thanks to Amazon bedrock, Alexa+ has access to this broad range of state of the art LLMs. Alexa, plus, is founded on a model agnostic system, so it’s always using the best model for any given task, and it gets the best out of each of them,” Mr Rauch said.

Alexa+ will be launched in the US initially from next month, and the rest of the world, including Australia, from later this year. Australian pricing is yet to be determined.

The author travelled to New York as a guest of Amazon