Amazon’s Alexa is undergoing its most significant transformation since its debut, as the company begins rolling out Alexa+, a next-generation voice assistant powered by advanced agentic artificial intelligence. More than a simple upgrade, Alexa+ signals Amazon’s ambition to move beyond reactive voice commands and toward an assistant capable of reasoning, coordinating, and completing complex real-world tasks through natural conversation.
For years, Alexa users have grown accustomed to asking their devices to play music, check the weather, or turn off the lights. Alexa+ aims to change that relationship entirely. Instead of responding only to direct instructions, the assistant can now understand context, plan actions, and carry tasks through to completion—from booking rides and reserving dinner tables to managing household routines without repeated prompts.
“We’re at the beginning of a long journey,” said Nedim Fresko, Vice President of Technology for Alexa. “Now that we have this strong technology foundation, this is just the start of what we want to do. Alexa+ will grow and evolve with our customers.”
Reimagining Alexa in the Age of Generative AI
The evolution of Alexa+ was neither quick nor simple. According to Amazon executives, the challenge was not merely adding generative AI but rearchitecting Alexa at global scale—while maintaining reliability for hundreds of millions of users.
“We couldn’t just rebuild Alexa from scratch with AI at the center,” explained Mara Segel, Director of Product Management for Alexa. “We had to preserve the trust customers already have, while introducing emerging technologies that elevate the assistant to an entirely new level.”
To do so, Amazon brought together teams from Devices & Services, Artificial General Intelligence (AGI), and Amazon Web Services (AWS). The result is a system capable of orchestrating dozens of large language models (LLMs) while seamlessly interacting with tens of thousands of services and devices.
“Reinventing Alexa involves rethinking everything—conversation, reasoning, and execution—without leaving any customer behind,” said Daniel Rausch, Vice President of Alexa and Echo. “And we have to do it at massive scale.”
APIs, Scale, and the First Major Breakthrough
A critical milestone came with Amazon’s ability to connect AI models to real-world services through APIs at unprecedented scale. Built on Amazon Bedrock, the company’s generative AI platform, Alexa+ can already coordinate multiple models to perform tasks across smart homes, ride-sharing, dining, and local services.
Users can now say things like “I’m chilly”, and Alexa understands the implied action—turning up the heat. Follow it with “Dim the lights”, and the assistant instantly adapts to the new context without restarting the conversation.
Amazon is further expanding this ecosystem with its Actions SDK, a developer tool that allows partners to integrate their existing APIs more easily. By mapping services into standardized categories—such as reservations or food delivery—Alexa+ can scale rapidly without having to “learn” each new system from scratch.
When APIs Fall Short, Agentic AI Steps In
Not every service provider has modern APIs, and this gap historically led to Alexa’s most frustrating response: “I’m sorry, I can’t help with that.” Alexa+ tackles this limitation through Nova Act, a new agentic AI model with advanced visual reasoning.
Unlike traditional assistants, Nova Act can navigate websites visually, much like a human—recognizing buttons, reading forms, handling pop-ups, and completing tasks securely in a browser session.
“Nova Act can handle any website, modern or old,” said Michael Giannangeli, Head of Product for Agentic AI at Amazon Nova. “It understands layouts and takes action just like a person would.”
For example, while Alexa could previously find a local plumber, booking the service still required manual effort. With Nova Act, Alexa+ can now complete the booking end-to-end, bridging the gap between digital intent and real-world execution.
The Power of Multi-Model Orchestration
At the core of Alexa+ lies a multi-model orchestration system, capable of routing each request to the most suitable AI model in real time. Some models specialize in fast, low-latency conversation, while others handle complex agentic tasks such as coordinating multiple services or navigating the web.
This orchestration ensures speed without sacrificing accuracy. “We have models optimized for every use case,” Giannangeli said. “From deep reasoning to lightning-fast responses, the system dynamically selects the best path.”
A Future of Ambient Assistance
For users, the promise of Alexa+ is simple: tasks that start and finish without friction. Ordering sushi, scheduling repairs, managing calendars, or setting up daily routines can now happen in a single, fluid conversation.
“This is only the beginning,” Giannangeli added. “These models will get faster, smarter, and more capable—and that will translate directly into more value for customers.”
Amazon envisions Alexa+ as a foundation for “ambient assistance”—technology that fades into the background while quietly supporting everyday life. Still in early access, Alexa+ is expected to evolve rapidly as more partners and AI agents join the ecosystem.
“What excites me most,” said Rausch, “is that we finally have the systems under one roof—AWS, Bedrock, Nova—that can deliver on a vision as big as Alexa Plus.”
