The real value of digital assistants isn’t automation – it’s augmentation

The conversation around digital assistants tends to gravitate around automation (AI bots doing the things we don’t want to do). Yet, at Claro we believe the real value will come from augmentation: digital coaches rather than digital butlers.

Sorry, who’s calling again?

The highlight of the last Google IO conference was the unveiling of Google Duplex, Google’s AI powered voice digital assistant.

In the demo, a woman calls a hair salon to make an appointment. After some back and forth small talk, she gets to the point and works out the best time for her hair appointment. If you haven’t watched the demo you should – you will see that it’s basically impossible to know that the woman calling is not human.

Last month, Google Duplex was rolled out as part of Google Assistant in 43 US states [1]. What was a futuristic innovation less than a year ago is now readily available to anyone with an Android phone. Yet Google Duplex isn’t particularly intelligent – its function is limited to the dull and admittedly simple task of making reservations [2].

So why the commotion? People’s bewilderment wasn’t about what Google Duplex did (make a reservation). It was about how it did it (brilliantly mimicking a human conversation – to the point it became impossible to know it is not “me” making the call).

This, in turn, highlights a number of fundamental questions: what is the relationship between Google Duplex and its “owner”? Is Google Duplex an external agent acting on our behalf – a kind of butler? Or is it rather a digital extension of ourselves? Who, in this instance, is actually making the reservation?

This question, while benign when the assistant’s role is limited to making an appointment, dramatically grows in significance once we start imagining what this new type of entity will soon be able to accomplish.

The advent of the Digital Selves

Everything we do today leaves a digital trace: meetings we set, emails we send, payments we make, content we consume, strings of text we tweet, pictures we post… the list goes on.

If Google Duplex can mine an agenda to find the best time to book an appointment, it takes little effort to imagine what it could do if it harnessed the whole trove of data in one’s Google ecosystem (agenda, email, payments, maps, searches, etc.).

Arguably, given the exponential growth of both the amount of data people produce and the ability of AIs to make sense of it, very soon this new type of digital assistant will be able to provide incredibly pertinent advice and recommendations with such accuracy, that people will happily let them act directly on their behalf.

To organise the countless applications of these Digital Selves, we can rank them on a continuum that ranges from automation (taking over tasks people don’t want to do – that’s the Google Duplex use case) to augmentation (enabling people to be better at what they do).

From one end of the spectrum to the other they will, amongst other things: organise your agenda, pay your bills, buy groceries, pick the best mobile or internet plans, suggest entertainment and leisure activities, plan your holidays, organise your finances, prepare you before meetings, coach you to gain new skills, suggest your next career move, and maybe even give you insights on your emotional wellbeing.

While the conversation on AI tends to gravitate towards automation, the tasks that can be automated are by definition those with the lower added value. This is why we believe the real transformation will actually come from augmentation [3]

If Google Duplex is a butler, we see the Digital Self as a coach: an entity that helps you know yourself better and gives you advice on how to be more efficient, how to grow, how to become a better version of yourself. In the words of Michael Schrage, research fellow at the MIT: “I want people to also think of AI as Augmented Introspection”.

Augmentation in the context of the workplace

While this will impact all aspects of people’s daily life, it foremost has the potential to completely transform the workplace. What happens when each employee has a ubiquitous personal coach, advising them on the best way to be more productive, more efficient, more sociable, or more creative? What is the impact on individuals, teams and organisations? More pressingly, what can your organisation do today to prepare for the advent of this new paradigm?

We will further explore these questions in a follow-up article which draws from our conversation with Michael Schrage, expert on the topic and research fellow at the MIT Sloan School’s Centre for Digital Business.

We can help

If you are interested in understanding how to best leverage artificial intelligence to augment human intelligence in the context of the workplace, contact us at; and stay tuned for our coming articles.

April 5th, 2019





[3] Of course, designing and developing solutions that can do a specific task instead of humans (i.e. automation) is relatively easy. As the solution is fairly self-contained, you have to conceive only a limited number of fairly simple human-machine interactions. On the contrary, designing a solution that provides inputs to people so they can do something better (i.e. augmentation) is incredibly more complex. At minimum it requires: (i) a fine understanding of the process at play (what are the different sub routines the human has to follow), (ii) a way to evaluate the outcomes of the overall process, (iii) knowledge that once transmitted to the human would allow him to be more efficient, or increase the quality of its output, (iv) an interface that would allow the human to take into account the solutions’ inputs without disrupting his workflow.

[4] Image from the promotional material of the movie Computer Chess,