Ten years after Amazon debuted its virtual assistant, Alexa is finally being upgraded for the generative AI age.
The new Alexa+, unveiled at a splashy event in New York City on Wednesday, transforms the familiar voice-activated chatbot into an action-oriented, interconnected service. Instead of simply telling you the weather or playing a song, the new Alexa can manage and coordinate a string of activities: booking a table at a restaurant while scheduling an Uber pickup for your spouse and texting the babysitter, for example.
Amazon CEO Andy Jassy helped unveil the new Alexa, describing Amazon’s mission of “using this new generative-AI revolution” to improve Alexa. “Our team has been working on this for several months. I’ve had chance to use it for the last several weeks. It is really remarkable,” Jassy said at the event.
The launch of the new Alexa, which will be available to limited customers in March, can’t come soon enough for Amazon, which has fallen behind amid a new crop of AI services such as OpenAI’s ChatGPT and Google’s Gemini. Amazon’s effort to catch up and revamp Alexa has been beset by delays, with the company initially teasing the new Alexa at an event more than a year ago.
In a sign of Amazon’s confidence in its long-awaited Alexa revamp, the company plans to charge consumers $20 a month for access to the service, though it will be free for Amazon Prime members. It works across Alexa devices, Amazon smart home devices, a new Alexa.com website, and a new Alexa app.
Wednesday’s event, kicked off by Jassy and led by Amazon devices head Panos Panay, showed off an Alexa experience that uses a variety of in-house and third-party AI tools, which the company believes will vault Alexa to the front of the pack. Among the state-of-the-art AI large language models powering Alexa+ are Amazon’s own Nova models as well as LLMs from Anthropic, the AI startup in which Amazon has invested $8 billion, and partnerships with a variety of partners including TicketMaster, GrubHub, Uber, and Amazon-owned Whole Foods.
Daniel Rausch, vice president of Alexa and Echo, emphasized the AI agent capabilities of Alexa+, calling it the “largest integration of services LLMs and agentic capabilities we know of anywhere,” and called out Anthropic’s chief product officer Mike Krieger for his work on the new Alexa.
The mix of LLMS appears to be a shift from Amazon’s strategy in September 2023, when it teased a revamped Alexa based on a new, custom-built large language model (LLM). At the time, Rohit Prasad, Amazon’s SVP and head of its AGI team, called the new Alexa a “super agent.” But according to Fortune reporting, former employees said the Alexa LLM was far from state-of-the-art. As OpenAI jumped ahead with ChatGPT’s Advanced Voice Mode in 2024, and Google followed with its Gemini voice mode, it seemed like Amazon was poised to lose its chance to keep pace with its competitors and bring Alexa into AI’s future.
Still, Amazon’s AI-powered Alexa has had many stops and starts over the past two years. And many questions still remain. In advance of today’s event, the Washington Post reported delays in making the new Alexa+ public because of inaccurate answers to test questions. There were also no comments from Amazon about data privacy or safety.
Shares of Amazon were up roughly 1.2% in midday trading on Wednesday, at $215.37.