Amazon Rebuilds Alexa into Alexa+ with Hundreds of AI Tools
–
Daniel Rausch, vice president of Alexa and Echo, is overseeing a major transition at Amazon. More than a decade since the debut of Alexa, his mandate is to build a fresh voice assistant powered by large language models. He calls this new release, dubbed Alexa+, "a complete rebuild of the architecture." Under Rausch’s direction, engineers have assembled fresh pipelines for model training, real-time inference and end-user interactions.
To build the system, Amazon engineers leaned heavily on generative AI at every stage. They used LLMs to draft snippets of Python and Java code, sketch out interaction flows and draft test scripts. "The rate with which we're using AI tooling across the build process is pretty staggering," Rausch says. And yes, that includes generating parts of the code. Teams worked with iterative prompts, fine-tuning outputs and folding those results into code reviews. Each iteration ran through automated checks before landing in internal repositories.
Generative AI also reshaped Alexa’s quality assurance. Engineers set up reinforcement learning loops where the assistant produced multiple responses, then scoured them with "a large language model as a judge on answers." That approach let the system pick the most suitable version between two Alexa+ outputs. After the AI selection, human reviewers would audit edge cases and flag errors. This method reduced backlog of manual quality tests and helped teams focus on refining dialogue and handling unusual requests.
That AI-driven pipeline cut cycle times for new features. Development squads could generate a proposal, receive model-guided feedback and push updates within days, rather than weeks. "People are getting the leverage and can move faster, better through AI tooling," Rausch explains. In parallel, the group has tracked metrics on response relevance, execution accuracy and user satisfaction. Early indicators suggest iteration velocity has more than doubled compared with previous Alexa releases.
As Amazon expands AI tooling, its workforce needs are shifting. A memo from CEO Andy Jassy this week spelled out the coming realignment. He wrote, "We will need fewer people doing some of the jobs that are being done today, and more people doing other types of jobs." He followed with, "It’s hard to know exactly where this nets out over time, but in the next few years, we expect that this will reduce our total corporate workforce as we get efficiency gains from using AI extensively across the company." Teams will retrain or move into AI oversight, safety and service optimization roles.
For now, Rausch is focused on expanding Alexa+ to as many users as possible. "We really didn't want to leave customers behind in any way," he says. "And that means hundreds of millions of different devices that you have to support." Beyond Echo speakers, teams customized the assistant for Fire TVs, Ring doorbells, third-party smart displays, Android phones and car dashboards. That breadth presents a challenge, given diverse OS builds and hardware limits, but Amazon’s AI pipeline handles most of the optimization automatically.
Alexa+ greets users with fluid conversation that remembers prior exchanges and suggests follow-up actions. It can book concert tickets, reorder weekly groceries and adjust smart-home settings on command. Each request taps into Amazon’s storefront, calendar and media libraries, reducing extra steps. At demo events, engineers showed how the assistant could summarize long emails or check real-time weather by name. That kind of multitasking support aims to cut friction and let people rely more on voice interactions.
Amazon unveiled Alexa+ at its February hardware showcase, then shipped an early preview in March with core features still in development. Testers saw a trimmed feature set that lacked calendar sync, advanced translation and cross-device memory. In a few weeks, Amazon pushed three incremental updates, adding voice prompts and security filters. Roughly one million people have joined the preview program so far, a small fraction of Amazon’s overall base. A wider release across compatible devices is set to begin later this summer.
The field is crowded. OpenAI’s Advanced Voice Mode arrived in 2024 and drew praise for dynamic tone shifts and clearer enunciation. Apple showed off a Siri upgrade at last year’s Worldwide Developers Conference, touting contextual awareness, persistent memory and deeper app integration. Apple has kept its new assistant under wraps since the demo. It is slated for release next year, according to sources familiar with the company’s timeline.
WIRED requested early access to Alexa+ for a hands-on review but was declined. As availability expands, WIRED plans to test the assistant live, evaluate its real-world performance and share user-facing impressions with readers.
