Amazon is trying to turn Alexa from a familiar voice assistant into something more capable: an assistant that can act for users, not only respond to them. During Amazon's Q3 2024 earnings call, CEO Andy Jassy pointed to a future version of Alexa shaped by generative AI and new foundation models.
The stakes are large because Alexa is already in over a half billion devices worldwide. Yet the product has not become a meaningful contributor to Amazon's bottom line, and the company is still working through the technical and business questions behind its next version.
Jassy's clearest hint yet
On Thursday, Jassy described where Amazon sees assistants and generative AI applications going next. His point was not simply that assistants will become better at producing answers. He said they will become better at doing things.
"I think that the next generation of these assistants and generative AI applications will be better at not just answering questions and summarizing, indexing, and aggregating data, but also taking actions,"
He then connected that shift directly to Alexa: "And you can imagine us being pretty good at that with Alexa." That phrasing matters because it places Alexa inside the broader move toward agentic AI, where software is expected to carry out tasks on a user's behalf.
For Alexa, that would be a meaningful change in role. The assistant has long been associated with voice commands, smart home control, information lookup, and everyday requests. Jassy's comments suggest Amazon wants the next version to become more useful across actions, not just answers.
The Alexa rebuild centers on new foundation models
Jassy also said Amazon continues to "re-architect the brain" of Alexa with "a new set of foundation models." He said the company plans to reveal those models "in the near future." The source does not give a launch date for that reveal, but the phrasing indicates the work is still active and central to the redesign.
Amazon first announced that it would revamp Alexa with generative AI technologies in 2023. Since then, the effort has appeared more complex than a routine product refresh. The article says Amazon is reported to be replacing its own Alexa-powering models with Anthropic's after encountering technical challenges.
That reported shift is notable because Amazon is a major investor in Anthropic. The source does not state the full scope of the model replacement, but it frames the move as part of Amazon's attempt to improve the Alexa redesign after internal technical issues.
In plain terms, the challenge is that a more agentic Alexa has to be both smarter and dependable. If an assistant is only answering a question, a slow or weak answer is frustrating. If it is taking action for a user, reliability becomes even more important.
Reports point to delays and performance issues
The unreleased upgraded assistant has reportedly struggled during the redesign process. At one point, it reportedly had trouble turning on smart lights and took up to six seconds to respond to queries. Those examples are limited, but they show why the redesign may be difficult: Alexa's next version has to handle generative AI while still doing basic assistant work quickly.
The new Alexa is code-named "Remarkable Alexa" internally. Reports have indicated that it would cost $5 to $10 per month and sit alongside a less capable free plan. That would make the upgraded assistant not only a technical change, but also a business model change for a product many users have experienced as part of their device purchase.
Some reports indicated the new Alexa would arrive in October. The article says it is seemingly suffering delays, and Bloomberg reports the timeline has slipped into 2025. That delay gives Amazon more time to refine the assistant, but it also raises expectations for what the paid version would need to deliver.
- Technical risk: the assistant must respond quickly and handle ordinary tasks reliably.
- Product risk: a paid Alexa has to feel meaningfully better than the free plan.
- Business risk: Amazon needs Alexa to do more for the devices business than it has so far.
The business pressure behind a smarter Alexa
Alexa's reach is not the problem. The assistant is already in over a half billion devices worldwide. The harder issue is whether that reach can become a stronger business for Amazon.
The article says Alexa has not contributed meaningfully to Amazon's bottom line. It also says Amazon has lost tens of billions of dollars in its devices business since 2017, according to The Wall Street Journal. That context helps explain why a more capable version of Alexa may be tied to a paid plan.
A free plan and a more capable paid version would let Amazon keep Alexa available broadly while asking users to pay for advanced features. The source does not list what those paid features would include. Still, Jassy's comments point toward action-taking ability as the broader direction.
That makes "Remarkable Alexa" a test of whether generative AI can change the value of a voice assistant. If the new assistant can only summarize, answer, and aggregate information, it may not feel dramatically different from other AI tools. If it can reliably take actions, it could give Alexa a clearer role inside Amazon's device ecosystem.
What to watch next
The next major signal will be Amazon's reveal of the new foundation models behind Alexa. Jassy said that reveal is planned for "in the near future," but the reported product timeline now points into 2025.
Until then, the central question is whether Amazon can make agentic Alexa reliable enough for everyday use. The company has the device footprint, the consumer brand, and a clear reason to improve the business. The redesign now has to prove that Alexa can move beyond answering questions and become useful enough to justify a more ambitious role.