From the Fall of Magic shall rise the Mini-Butlers
On January 4th, 2016, Magic, the much-hyped virtual butler application, announced they were going to charge users $100/hour to use their app. It remains to be seen whether a free model will continue to exist, but this change in attitude reflects the primary concern many have had with virtual butlers — they are simply not scalable. However, from the ashes of Magic, we are likely to see a more successful model appear: company-backed mini-butlers.
Why Magic Worked…
In newer generations of consumers, two counterintuitive trends have appeared: a preference for passive communication, and a dislike of the infinite amount of choice the internet offered.
The easiest way to get a problem solved quickly (calling an electrician, arranging a night out with friends) is often active communication. A short phone call is sufficient for arranging things within minutes. However this approach is extremely high-involvement — something that the multi-tasking consumer dislikes. It is much lower involvement to set up a chat, where the user can check on answers and comment whenever they have 12 seconds to spare. It’s why so many meet-ups are arranged over text-based chat, even though a call would be much faster.
Meanwhile, the internet has showered us with so much choice that we have had to resort to specialized comparative engines to help us sort through it (Kayak, The Zebra, among many many others). More choice is not always better, as the famous Jam Study pointed out. For those unfamiliar with the story, the Jam Study showed that consumers were ten times more likely to purchase a jar of jam when presented with 6 options than they were when presented with 26. Too much choice requires far too much active effort.
Which is why Magic worked so well. Their solution was at the perfect intersection of these two consumer frustrations. Their virtual butler sorted through choice for them, while their interaction remained text-based and low involvement. It was the perfect idea.
…and why it is failing
In order to work though, Magic needed a real operator on the other side of the conversation, to sort through choice and handle the transaction. As it grew, Magic needed more and more of these operators, needed them to be trained to be courteous, polite and helpful, and needed to retain them. That adds up to incredible labor costs, and makes scaling the business close to impossible.
Worse yet, live operators (and the turnover they would have), means that Magic wouldn’t be getting smarter with each demand. Particular consumer preferences might go unnoticed, trends of what was increasingly popular (and potential partnership opportunities) would get lost as clients interacted with different operators.
The only way to scale such a business is AI. Not only would labor costs almost completely disappear, the AI could learn from every interaction, understanding each consumer and trends perfectly. Navid Hadsaad, the CEO of GoButler, Magic’s main competitor, agrees that what is needed is an AI that can understand and respond to natural language. However, an AI with the robust power to filter through tons of information and adapt its offer to a consumer’s reaction would not only be incredibly difficult and expensive to develop; it would be the next Google. Why use Google when you have a great AI that can search it for you? Its potential would be wasted within an app like Magic.
Magic and GoButler are aiming too big. The AI they would need to make their product scalable and sustainable is far too complex.
However, both these companies were onto something, and major companies could learn from it.
Perhaps instead of large, complex butlers, what consumers need are smaller, specialized butlers. Imagine having Kayak (flight bookings), Fandango (movie ticket bookings) and Eat24 (restaurants) as Whatsapp or Messenger contacts. Each runs a very simple AI linked to their existing database that acts as your mini-butler for that particular service.
It would look like this:
The beauty of this approach is that it gives companies yet another way to communicate in a personal way with consumers (real communication, not just pushing content), learn more about each of their consumers’ preferences, and make do with a simplified AI: they already have the databases, and we don’t need natural language response — keywords are more than sufficient (Star Wars Episode 7, after 8, user’s location).
Finally, it provides a simpler approach than apps. Users can stay within the main app they use, and only need to use Fandango’s app if they are looking for in-depth information (and even that can be done with a mini-butler eventually).
Magic is gone, but from its ashes will likely grow an army of mini-butlers.
The Next Big Thing
The real winner? Any company that can come up with a text-chat AI that can be customized for different types of products (flights, movie tickets) and linked to company databases. The real money in virtual butlers is not with consumers — it’s in B2B, selling them to businesses that will use them with the consumers they already have.