The evolution of the digital assistant
Defining the ASA (autonomous self-agent)
For decades, fiction writers and scientists have been working on personal assistants. The original vision was an anthropomorphic, mechanical machine that walked and talked like a human and was little more than a glorified butler.
Since the digital era, we’ve moved away from that model to something that is more ephemeral. Usually a fancy animation on a screen accompanied by a soothing but still synthesized voice that patiently waits for you to command it to do something.
In short, we went from a metallic bipedal butler to a disembodied butler. Delightfully boring.
Not too long ago, thanks to the rapid advancement of machine learning and deep neural networks, we’ve been able to give machines the power to learn and recognize things. By simply showing an algorithm thousands of pictures of different cats, Google was able to make the algorithm understand a picture of a cat it had never seen before. This is similar to the process we humans learn by, except because it’s a computer that doesn’t get tired, what takes us days — or weeks — as a child, only takes the machine a couple of minutes to learn.
It’s powerful, and in the context of a digital assistant, it gives the butler a lot more freedom to understand you and give you better suggestions for flights and hotel bookings. Better, but still boring.
The ideal assistant would be completely autonomous. It doesn’t need to wait for you to command it to do something that would be beneficial for you. For instance, a digital butler shouldn’t wait for you to tell it that you are thirsty and that you would like a lemonade. Through your previous actions and biometric sensor data, the butler should command a drone to drop a lemonade into your hand at just the right moment. This level of autonomy is crucial for the enjoyable use of digital assistants.
The next step, is to make it highly personalized. This is truly where a Self-Agent (SA) and a Digital Assistant (DA) diverge. While both can be autonomous, the lemonade ordering computer explored above is a great example of an Autonomous Digital Assistant (ADA). The SA on the other hand is another beast.
With the growing use of social media and the increased amount of time spent managing it, there is need for a new kind of assistant. If machines and DA’s were designed based on the functions of plumbers and maids, SA’s such as Doliio are designed based on the functions an ambassador would have.
An ambassador best understands who they represent and does so well to manage and maintain the host’s relationships. With the newest technology in learning and computations however, we’re able to not only have an ambassador represent you, the ambassador will learn to be the real you depending on who they are communicating with. Enter the Self-Agent.
Not to be confused with self-agency, although one might argue a self-agent has that, the etymology stems from it being an agent of the self — more specifically yourself.
Having defined this much, we can put it into the context of Doliio now. If Doliio was a mere SA, you would have to continuously tell it to take actions such as congratulating your best friend on his baby by requesting that it comments on and likes the photo on Facebook. While this might be convenient, in that it removes the tax of actually going to the website and performing those actions yourself, it’s still not enough. Doliio is able to see that your friend got a baby, look at how you think of babies and that friend and make an authentic and appropriate comment on it. Then, in the report on the following day, Doliio will tell you to mention that when you see him in person. This is true autonomy, and this is what makes Doliio an Autonomous Self-Agent (ASA).
We have seen other SA’s around the web but Doliio is the only true ASA to date. While we would love to be the only ones in this space, it is inevitable that more ASA’s will come. We hope that you sign up for the Doliio beta at doli.io and help us develop the best and most authentic ASA around.