Amazon is set to release its long-awaited – and delayed – Alexa generative Artificial Intelligence (AI)  voice service, said three people familiar with the matter, and has scheduled a press event for later this month to preview it.

Once released, it would mark the most significant upgrade to the product since its initial introduction accelerated a wave of digital assistants more than a decade ago.

Amazon on Wednesday sent press invites to an event to be held on February 26 in New York featuring the head of its devices and services team, Panos Panay. A spokesperson said the event is Alexa-focused, while declining to elaborate. 

The new generative AI-powered Alexa represents at once a huge opportunity for Amazon, which counts more than half a billion Alexa-enabled devices in the market, and a tremendous risk. Amazon is hoping the revamp, designed to be able to converse with users, can convert some of its hundreds of millions of users into paying customers in an effort to generate a return for the unprofitable business. 

The AI service will be able to respond to multiple prompts in sequence and, company executives have said, even act as an “agent” on behalf of users by taking actions for them without their direct involvement. That contrasts with the current iteration, which generally handles only a single request at a time.

Executives have scheduled a meeting, known as a “Go/No-go,” for February 14. There they will make a final decision on the “street readiness” of Alexa’s generative AI revamp, according to the people and an internal planning document seen by Reuters. 

Alexa’s revamp carries with it all the challenges inherent in now-familiar generative AI chatbots from OpenAI, Alphabet and others including the possibility of fabricated answers, known as hallucinations. With access to Alexa available in cars, televisions, thermostats and mobile phones, it could become an essential daily tool for scheduling and even shopping.

Initially, Amazon plans to roll out the new Alexa service to a limited number of users and will not charge for it, the people said, though it has considered a $5 (roughly Rs. 437) to $10 (roughly Rs. 875) monthly fee. The company will also continue to offer what it is calling “Classic Alexa,” the version broadly available today for free. One of the people said Amazon has discontinued adding new offerings to Classic Alexa. 

Bezos’ Vision

While Apple’s Siri voice assistant preceded Alexa’s 2014 release by three years, the Amazon service supercharged the acceptance of voice assistants. But for many people, Alexa is now used for little more than kitchen timers and weather updates due to its lack of significant overhauls in the last few years.  

Alexa is the brainchild of Amazon founder Jeff Bezos, who envisioned a service that would resemble the voice-activated computers on TV’s “Star Trek.” The hope was that once perfected, users would turn to the voice assistant for hundreds of everyday tasks like turning on lights, preheating the oven, accessing the internet, playing music, writing emails and summoning taxis. 

“Someday in the future – that might be years or decades away – it could answer everything that you would ever ask it,” Amazon’s then-chief of devices, Dave Limp, said nearly a decade ago.

With those weighty expectations, the move to upgrade Alexa has suffered delays over concerns around the quality and speed of its responses, people familiar with the matter have told Reuters. Amazon dubbed the new service “Banyan” internally, as well as “Remarkable Alexa,” though it was not immediately clear if the Seattle company planned on using either as a new product name.

In a January Financial Times interview, Amazon executive Rohit Prasad acknowledged some of the obstacles in developing what is effectively an entirely new service, including the work to eliminate hallucinations.

Analysts at Bank of America estimate Amazon could generate $600 million (roughly Rs. 5,251 crore) annually if 10 percent of active users, which it estimates at around 100 million devices, pay $5 (roughly Rs. 437) per month for the service. 

Customer Preferences

The new Alexa is designed to remember customer preferences to help make music or restaurant recommendations and allow users to give multiple prompts in a single session, so that, say, one could later modify a hamburger order before it is sent out for delivery, the people said.

It will be compatible with existing Alexa devices, said the people.

Amazon will rely, at least in part, on AI software from startup Anthropic – in which it has invested $8 billion (roughly Rs. 70,023 crore) – to underpin the AI in Alexa, people familiar with the matter told Reuters.

In September 2023, Amazon demonstrated an early version of the souped-up Alexa, but concerns over quality caused Amazon to repeatedly push back its launch date and that rendition was never released to the public.

© Thomson Reuters 2025

(This story has not been edited by NDTV staff and is auto-generated from a syndicated feed.)

For most of human history, working out has been a community activity, either outdoors or in designated spaces such as gyms and sports centres. However, with the rise of the Internet and other connected technologies, businesses have figured out ways to turn it into an “at-home” experience. This has its advantages as well. For instance, the average office-goer who spends a good chunk of his weekdays commuting and at work might find exercising at home more convenient than going to the gym after a long day.

Another group of people who benefit from this are the elderly who might not enjoy travelling for exercise due to mobility issues. This is what led to the arrival of various home workout programmes, video tutorials, and online classes. While it did address the problem of accessibility to some extent, one area that was still largely unaddressed was personalisation. But not for long.

Role of AI in Home Workouts

The artificial intelligence (AI) discipline might thank the wave of generative AI that started in late 2022 for mainstream attention, but its impact on the lives of people was felt much before that. From smartphones to search engines and from Microsoft’s Clippy to Amazon’s website, AI technologies such as predictive analysis, rule-based AI, and machine learning (ML) have always been part of popular software.

AI also made its way to the home workout space slowly. Products such as Fitbit and Apple Watch with health and fitness sensors gave people insightful data about their daily activity level. These devices also allowed users to make adjustments to get more out of their workouts opening the path for personalisation.

While the pieces were being put together, the home workout space was missing one key ingredient — expert advice on type of workouts, posture, and an option to build a personalised routine. However, during the COVID-19 pandemic, when the majority of the population was stuck indoors for months, businesses cracked the code.

Carol Bike 2.0, OxeFit’s XP1, and Vitruvian Trainer+ were some of the home gym equipment that sought to bring AI personalisation to people. But they were largely catering to the Western countries and India was missing out. Sensing an opportunity to address this gap, Portl, a Hyderabad-based startup entered the space with its smart mirror-based AI home workout system dubbed Portl Studio.

Gadgets 360 spoke with Indraneel Gupta, Founder and CEO of Portl to understand the tech behind Portl Studio and how the startup is leveraging AI to offer users a personalised experience.

The Vision Behind Portl Studio

Founded in 2021, Portl aimed to solve the same personalisation problem for those who prefer working out at home. Gupta said, “One common issue that persists even today is that gyms and fitness centres tend to follow a very cookie-cutter approach to fitness that doesn’t apply to everyone.”

Portl Studio
Photo Credit: Portl

 

The company focused on the demographics of people aged 35 and above and found that fitness was as much a convenience issue as it was an engagement issue for them. Their solution? A device with a large smart mirror that can be spotted from a distance — which also doubles up as a display that shows personalised workout routines and follow-along guides to help them get the right motivation. But, how does it all work together?

The Portl Studio Tech Stack

Portl Studio is a 32kg device with dimensions of 5.8 x 2 x 0.1 feet. It is a fairly large and heavy device that requires experts to properly install it to either a free wall inside a house or on a stand. While this would make it a piece of unfeasible equipment for some, it would also be an ideal setup for those whose main struggle is to get out of the house and go to the gym after a long day. For those people, the device tries to fit in all the requirements that could eliminate the need for a social space.

The Portl Studio’s mirror features several sensors that track the user’s posture in real-time. It also comes equipped with biosensors, similar to a smartwatch or a fitness band, that can scan health vitals. The device also implements several cameras to properly assess all of these parameters.

There are two components that enable the data processing as well as allow the user to interact with the system. The first is the system-on-chip. For that, Portl uses Snapdragon chipsets to power the device. It handles all the computational power that the device would need.

The second part is the touch screen, which is a multi-point touch display. The company said that the display is coated with a nano protection layer that adds durability and protects against breakages and damage. The device also connects with Wi-Fi and Bluetooth to let the user either connect their earphones or stream their playlist from Spotify.

“So the idea has always been, how do we create a scenario where the program constantly adapts to the user’s minimum level of performance? We designed a system that looks at variable data, the sleep routine, recovery rate, and more. All of these factors go into the next workout routine that they follow,” Gupta explained.

Portl’s solution was an intelligent machine that prioritises real-time data to process what the user needs in terms of workout intensity. Interestingly, this heavy lifting was done by the company’s AI processes.

AI Behind the Portl Studio

Gupta explained that the entire data collection and AI processing occurs on-device and none of the personal user data ever leaves the system. To handle such complex tasks in real-time, the company included several AI-based processes.

Notably, no generative AI tools were used for the system. Instead, the company uses classical rule-based algorithms to both analyse and predict user behaviour. Gupta explained that the decision to opt for a rule-based algorithm instead of something more dynamic was because fitness science, kinesiology, and body mechanics are not constantly changing and are well-defined.

However, just like generative AI models, these AI systems are grounded at a first principles level to ensure that the system can gauge a base level of user performance. On top of that, the information provided by the user creates an extra layer of personalisation.

Portl Studio AI interface
Photo Credit: Portl

 

To explain how it works, Gupta gave an example, “During the onboarding, a user tells the system about their current lifestyle, injuries, health conditions, and more. All of these factors determine the kind of exercises and movements that will be used to create the workout plan. Then, once a user starts the plan, a 45-second health scan measures the core vitals of the user. This includes metrics such as blood pressure, heart rate, cardiac stress, etc.”

As per the Portl CEO, the scan then creates a baseline for the individual. Further, the cameras also collect data on the user’s functional movement screen, which includes balance, range of motion, and mobility. All of the data is used to understand the user’s starting point. From there, the user gets a workout plan that both pushes them and notes their limitations.

Portl Studio’s Use Case

The Portl Studio offers 15 different workout formats, with calisthenics as its core principle. This means this device does not support weight-based workouts and only focuses on cardio and bodyweight exercises. It also comes with various yoga and meditation courses. Additionally, there are skill-based exercises such as Muay Thai and Kickboxing.

At present, all 15 workout formats and guiding videos from experts are part of the device’s native offerings. This means the user does not need to pay subscription or add-on fees to access any of the content. As a result of all these offerings, the company claims it has more than 1,000 monthly active users despite a steep price of Rs. 1,25,000.

While Portl Studio offers a lot in terms of personalisation, ease of access, and choice of workout, the lack of strength-based workouts does create a significant gap. Gupta realises this, and that’s why the company recently launched a product that caters to this segment — the Ultragym.

Portl Ultragym

Ultragym is a smart fitness device that takes up 2.4 sqft of space and weighs 12kgs. It features a board, cables, accessories, and a bench. The company claims that users can perform 150 different workouts using this device. The cables do not use weight plates and instead use a motor to increase the resistance.

Portl Ultragym
Photo Credit: Portl

 

Each cable can be pushed to a maximum of 35kgs for a total of 70kgs of weight for a workout that requires both hands. The device also comes with a companion app that monitors and analyses workouts and creates personalised routines for users. Portl’s Ultragym is priced at Rs. 59,990.

On ensuring safety while users performed workouts with heavy weights, Gupta explained that the device has inbuilt safety features. If a user struggles with the weight or loses balance, the sensors can automatically reduce the weight to allow them to get out of that position safely.

The Future of Smart Home Workouts

Smart home workout technology has come a long way. However, there are still several gaps to be filled. First, personalised routines and fitness tracking have enabled users to make progress without leaving the comforts of home, however, existing devices are not portable enough to be carried everywhere. People who travel frequently will not find such gadgets very useful.

Second, these technologies are focused on basic fitness and do not have solutions for those who have competitive and advanced fitness goals. Further, most of these technologies do not emphasise the motivation element enough, which is a key factor in whether or not a user exercises consistently.

As companies experiment with innovation, these gaps are likely to be filled in the coming years. Till then, whether working out can become an at-home experience or if a social institution such as a gym or fitness centre is necessary, debate is up the air.

Apple is said to be working on a new product in the smart home product category. Bloomberg reporter Mark Gurman has mentioned that Apple is gearing up to unveil a HomePod-like smart home hub with an iPad-like screen this year. It is said to come with a touch display and could allow users to control smart home accessories and connected appliances. It is expected to come with a 7-inch display with support for Apple Intelligence features. 

Smart Home Hub Could Be Apple’s Most Significant Launch of 2025

In his latest Power On newsletter, Mark Gurman revealed that Apple will unveil a new HomePod-like smart home hub with a touchscreen later this year. It is claimed to be “Apple’s most significant release of the year,” as it represents the company’s “first step toward a bigger role in the smart home.”

Gurman notes that the smart home hub will be like a smaller and cheaper iPad that lets users control appliances, conduct FaceTime chats, and handle other tasks.

Meanwhile, Apple supply chain analyst Ming-Chi Kuo had stated earlier that the new HomePod with a 6-inch to 7-inch display will enter mass production in the second half of this year. It is said to pack an A18 chip and offer support for Apple Intelligence. If launched, the purported HomePod will compete against the likes of Amazon’s Echo Show and Google’s Nest Hub.

The upcoming model is tipped to carry a square display instead of a rectangular one. It could include a customisable lock screen with multiple clock faces. It is likely to include an in-built camera to support video conferencing apps like FaceTime. The camera may be able to recognise hand gestures.

As per Gurman, the price of Apple’s purported smart home hub will be $1,000 (roughly Rs. 83,740) or more. The company is reportedly working on a new operating system dubbed homeOS to run these devices and its existing smart home lineup including HomePod and HomePod mini. The basis for the OS is said to be tvOS.