Amazon Revamped Alexa with an Impressive Array of AI Technologies

Daniel Rausch, Amazon’s vice president of Alexa and Echo, is navigating a significant shift. Over a decade after the debut of Amazon’s Alexa, he’s responsible for developing a new iteration of the flagship voice assistant, one that utilizes large language models. As he stated in my conversation with him, this innovative assistant, referred to as Alexa+, represents “a complete rebuild of the architecture.”

How did his team tackle Amazon’s most extensive overhaul of its voice assistant? They leveraged AI to construct AI, naturally.

“The pace at which we’re utilizing AI tools throughout the development process is quite impressive,” Rausch remarks. Throughout the new Alexa’s creation, Amazon incorporated AI at every stage of development, including generating some of the code itself.

The Alexa team also integrated generative AI into the testing phase. Engineers employed “a large language model as a judge on answers” during reinforcement learning processes, where the AI determined the best responses from two outputs of Alexa+.

“People are gaining leverage and can operate more quickly and effectively through AI tools,” Rausch notes. Amazon’s emphasis on internal generative AI is part of a broader transformation in the software engineering landscape, as new tools, like Anysphere’s Cursor, alter job routines and expectations.

If such AI-centric workflows are proven to be extremely efficient, the role of engineers will undergo a fundamental shift. “We will require fewer individuals for some of the tasks currently being performed, and more individuals for different types of work,” stated Amazon CEO Andy Jassy in a memo to staff this week. “While it’s tough to predict the long-term impact, we anticipate that in the coming years, this will decrease our overall corporate workforce as we achieve efficiency gains through extensive AI use across the company.”

Currently, Rausch’s main focus is on making the generative AI version of Alexa available to more Amazon users. “We really didn’t want to leave any customers behind,” he explains. “This entails supporting hundreds of millions of various devices.”

The new Alexa+ engages in a more conversational style with users, offering a personalized experience that recalls preferences and handles online tasks you assign, such as searching for concert tickets or purchasing groceries.

Amazon unveiled Alexa+ during a company event in February, providing early access to a select group of public users in March, albeit without the full range of anticipated features. Currently, the company reports that over a million people have access to the updated voice assistant, a fraction of the potential user base; ultimately, hundreds of millions of Alexa users will gain access to this AI functionality, with a broader rollout potentially planned for later this summer.

As Amazon develops a more versatile voice assistant, it faces competition from various contenders. OpenAI’s Advanced Voice Mode, introduced in 2024, became popular due to its engaging AI voice. Additionally, Apple revealed a complete overhaul of its built-in voice assistant, Siri, at last year’s developer conference, featuring many contextual and personalization elements similar to those found in Alexa+. Although Apple has yet to launch the revamped Siri, it is anticipated to debut sometime next year.

Amazon chose not to provide WIRED with early access to Alexa+ for hands-on (or voice-on?) testing, and the new assistant has not yet appeared in my personal Amazon account. Following a similar path to our approach with OpenAI’s Advanced Voice Mode, which launched last year, WIRED intends to test Alexa+ and offer experiential insights for readers as it becomes more accessible.