Artificial intelligence: How companies can capitalize on the coming boom in 'smart assistants'

The Oura Ring is a wearable device that tracks our sleep, monitors our hearts and breathing rates, and tells us how many calories we’ve burned every minute. It can also predict health events, discerning subtle changes in body temperature that presage, say, the start of a woman’s menstrual cycle. And, in a recent university study, it was able to detect disease, diagnosing COVID-19 among healthcare workers three days before any symptoms appeared.

The device is one of a growing number of smart assistants operating in our homes, cars, and workplaces, as well as on our bodies. Some, like Amazon’s content-curation algorithms, which generate product recommendations, are invisible helpers. Others, such as Google Nest, sit on our kitchen counters and nightstands to help us interact with and control our environments. Still others, like the Peloton bike and Samsung’s JetBot robot vacuum, help with specialized tasks.

Voice-enabled virtual agents are not limited to the pages of science fiction novels anymore: In five years, whether as wearables, appliances, devices, gadgets, or tools, they will be embedded in just about everything we use. Getting information or making a purchase will be as simple as expressing the thought out aloud. As our devices get to know us, absorbing data, discerning patterns, and gauging sentiment and context, they’ll anticipate our thoughts and needs.

Smart speakers will wake us at the optimal time based on our sleep rhythms and morning schedules. Smart mirrors will tell us if the blemish on our chin is acne, or something to call our dermatologist about. Home hubs will gauge whether our plants have enough water and if we’re low on milk, adding items to our grocery lists and moving up delivery dates as needed.

At work, A.I.-enabled voice assistants will transcribe notes and manage our calendars, and afterward, they’ll accompany us to the gym, where they’ll check our vitals and recommend workouts. Should our socks spring a hole, a quick word to our device will have a new pair on its way. These capabilities will create a marketplace whose size will be around $45 billion by 2027, upending commerce.

From lifestyle partner to industry disruptor

Smart assistants have become a fixture in our daily routines because of the convenience they enable. Opening the cupboard and rattling off a list of sundries for an A.I.-based device to purchase takes five minutes, whereas driving to the market, searching for items, and queuing to pay can take an hour. With the time saved, we can get work done, get a workout in, or simply rest or recreate—and our devices enable these activities, too! They are our curators, our coaches, and, as their knowledge of us grows, our confidants.

People will still surf the web and go to the store, but with smart assistants they won’t need to do so with the same frequency as they used to. As voice-enabled ecosystems expand, they’ll become the primary channel for shopping and engagement—and gatekeepers to customer relationships.

For all those reasons, voice-enabled A.I. agents pose a major disintermediation risk for consumer-facing companies. Value pools will shift accordingly. As in any platform economy, a few dominant players will exert a large gravitational pull. These companies will pick and choose which third-party offerings to recommend—including their own—and reap the cross-selling and upselling benefits that accrue.

Much the way that Apple’s app store reordered distribution, forcing brands to play by Apple’s rules, smart assistant platforms will do the same. Brands and retailers will have to fight harder to gain prominent placement, and they’ll have to pay a fee for every product they sell. Amazon, for example, pockets roughly one-third of the revenues from every third-party transaction on its website. The value shift will occur on a much larger scale in the smart-assistants marketplace. As we move from one-click to one-word shopping, purchases will surge, and so will the size of the addressable market.

However, two leaps will be required for smart assistants to deliver on their potential.

Where the smart money behind smart assistants will focus

The first challenge in the battleground is developing voice technology capable of engaging in human dialog and context-based conversation. As anyone who has wrestled with dictating a text message can testify, the speech recognition and synthesis capabilities of today’s digital assistants can be maddeningly inept. Language is stilted, language processing limited, and the assistants’ situational awareness is clumsy or non-existent. Addressing these issues will require major improvements in natural language processing, sentiment analysis, and empathetic A.I.

Many big tech players are bent on clearing these hurdles quickly, and they are well on their way to succeed. Over the last three years, the leaders have invested over $5 billion in natural language processing initiatives, and set up huge labs; Amazon alone has 10,000 employees working on its suite of smart assistants. The collective focus has knocked down barriers such as far-field speech recognition, making it possible for smart assistants to overcome background noises and make sense of what an individual is saying even if they’re in the next room.

The second challenge will be to deliver category-spanning recommendations personalized to each individual. Delivering such recommendations at scale isn’t just a technology problem; it involves building expertise and partnerships in multiple sectors and earning customer trust. Designers will need to assemble a choice architecture that reflects deep knowledge of the customer’s lifestyle, financial goals, preferred brands, and other information. Success isn’t just a matter of suggesting movie “A” over movie “B;” Netflix, and others, can do that today.

Machine learning intelligence must weigh the pros and cons of different choices across categories, and present us with recommendations based on a granular understanding of our needs and preferences. Instead of suggesting movie “A” or “B” at the cinema, for instance, a smart assistant might suggest that a customer watch a film at home for free by signing up for a trial subscription at a video streamer, use the savings to order take-out from a splashy new neighborhood restaurant or, if they’re feeling thrifty, use it to fund their metro card for the next week.

Maintaining significant stores of continually refreshed data is essential to deliver the breadth and the depth of recommendations customers need. Acquiring that data will require device-makers and content-producers to earn a customer’s trust, protect sensitive personal information, and deliver value in return for sharing data.

Rather than biting off everything at once, the leaders will solve the challenge progressively, specifically focusing on universes of needs such as the home, beauty and fashion, fitness, and work. The Oura Ring, for instance, focuses on health and wellness, the Peloton on fitness, and Google Nest on home security, temperature, and lighting. By laying down roots in highly personal spheres, the early movers can build intimacy and trust—critical to winning audiences for more ambitious offerings down the line. Having mastered one domain, leaders can extend their reach into other universes—in the same way that Amazon is doing in core commercial categories today.

In under five years’ time, we believe, a handful of smart assistant platforms backed by a cadre of companies will have acquired the voice technology and cross-cutting recommendation capabilities to serve the consumer. That’s why consumer-facing companies need to get into the race right away if they’re to stand a chance of being competitive down the road. Who the winning players will be isn’t clear, but the prospect of dominating is why many players are investing huge amounts in the business today.

The next five years promise to unleash a wave of innovation in the smart assistant space, which will be a winner-take-most market. With customer engagement likely to coalesce around a narrow set of channels, top management teams must start thinking about where and how they want to play, and what they must do to acquire the data, technology, and customer trust to realize their ambitions. Answering those questions thoughtfully will take time, but voice-enabled or not, the clock is already ticking.

Read more columns by François Candelon.

François Candelon is a managing director and senior partner at BCG, and global director of the BCG Henderson Institute. Karen Lellouche Tordjman is a managing director and partner at BCG, and fellow at the BCG Henderson Institute. Ariane Lafolie is a consultant at BCG and ambassador at the BCG Henderson Institute.