Article posted on the Official BAIA Blog on September 2007
Giorgio Manfredi is the founder and President of Kallideas a hi-tech Italian company. Kallideas approaches business in a dynamic and innovative way. Instead of waiting for support from Italian institutions, as many Italians companies like to do, this cutting-edge company is pro-actively building an international network of connections with top universities, research labs, and business associations like BAIA. The person that founded Kallideas and that continues to inspire it is Giorgio Manfredi. Here is a short interview with Giorgio.
Giorgio, can you tell us a bit about yourself and your company?
We were born at the end of 2002 with a mission to realize innovative projects in the field of business-users relations. Our reference customers were mainly telco and media agencies. Among our main customers there are Vodafone, Fastweb, Cartasì, Nokia, Dell, Motorola, etc. Our background is seated in the university and telecommunications fields. I represent its synthesis since I worked in both of them: I worked for many years as an associated researcher at the INFN (National Institute for Nuclear Physics). Afterwards I worked in several companies specializing in system architecture design for computer-telephony (CTI), Vodafone (then Omnitel) was one of them. For over a year, Kallideas has focused its efforts on the K-Human™ project, passing all the other assets to Key-One, one of the companies we control. Kallideas strengths are surely the people who work in it and the great work environment we were able to create: a real lab where technology, design, and creativity are well combined. Cultivating relations with the university research world is one of the main focuses of our assets, directed to constantly feed this “will to explore”. With the K-Human™ project we have gained a good visibility in Italy and Europe. The level of this project allowed us to get in touch with sector leaders in the field of technology for communication (obviously on this side of the Atlantic!) It is our intention to make the most of the opportunity we had to grow up and make Kallideas a presence also in the American market.
Kallideas is proposing an integrated approach to different channels and media (voice, video, text) of interaction with the end-user or customer. How do you manage to combine a general approach while exploiting the specific nature of each media?
There are two types of integration that need to be considered. The first, purely technical, relates to the choice of having a very same platform able to integrate different channels and media. We have developed a very flexible and modular software architecture that allows us to easily add any kind of input devices (such as web, telephones, touch screens, but we could just as easily add facial recognition devices which could give us information on user’s experience) and output devices without ever having to reconfigure the platform itself. Our approach therefore integrates various channels and media from a technical point of view. The second type of integration relates of course to the content that is being treated over different media and channels. Here too, it’s necessary to distinguish between knowledge acquisition (i.e. understanding what our K-Human™ virtual assistants have to do by discussing with our clients and analyzing their available data) and the interaction that is then presented to the final users. Knowledge acquisition is, of course, media independent: understanding and building model logics is related to client’s business, not to the channels the interaction with the final users will be done on. Therefore, our approach integrates procedures independently from the channels and media that will be used. On the contrary to technology and knowledge acquisition for which media and channels do not matter, we actually do have different interactions designed according to the nature of each media: on the phone, for instance, conversations have to be much shorter than what you can typically have over the web, and it would be suicidal not to differentiate such channels. Our whole policy is based on making machines come closer to the human ways of interacting, not the other way around. Our approach for designing interaction is therefore to use common logic but allow specific discussion flow decisions to be taken according to the media and channel the interaction is taking place on.
Another new idea of Kallideas is to integrate emotions in the interaction with the user. Diego Ventura, in an interview on this blog, said that is better not to emulate human emotion “to set the proper expectations for the customer”. What is your position on this topic?
This coin has two sides: one is of potential problems, the other of potential opportunities. Let me start from the potential problems. ACM Digital Library is packed of papers about human behavior emulation’s pros and cons. Most cons basically relate to a risk called “Uncanny Valley” that predicts a sort of repulsive response when robots are too much similar but distinguishable from a human being. In our point of view, we are still far from experiencing such a situation, and considering that Masahiro Mori (the scientist that first introduced the Uncanny Valley concept) focused his attention on robots, we believe that this risk is something related to a human like robot with a physical presence, not to an avatar that needs a screen to be visualized. No physical body, no Uncanny Valley. Now, about the opportunities. I see various different levels of opportunities for our K-Humans™ virtual assistants. First of all, people are connected as it has never before been experienced in the whole human history. Due to competition issues, companies need to overexpose themselves on different media and channels, and strive to face the raising number of interactions with people. Nowadays, companies need to talk with customers, partners, and third players more than ever. This is the time when companies can bet on intelligent and pleasant automatic systems, which are the only way to deal with the amount of contacts required by their overexposure. It is something already here, we are not talking about a nearby future. In the Long Tail era user interfaces are “conversational” interfaces. Think for a while on the blog revolution and how it changed the way in which people write and share text documents. Today, another revolution is taking place: social networks are taking the shapes of Second Life, Nintendo’s Wii and Mii, and PlayStation‘s “Home” virtual world. You may ask yourself a question: are you ready for a world in which everyone has avatars? Companies needs to talk the language their target users talk, and undoubtly virtual assistants are a relevant part of this new environment which is coming just one or two years from now. Our ‘second level’ of opportunities. This thought drives me to the last point. Humans (atoms) and machines (bytes) have experienced a deep gap in the past. In the near future we will find hundreds of ways to fill this gap. New hardware, new mobile devices, new interfaces: machines need human-like behaviors. We are producing culture, by developing technologies and 3D engines, by researching in the modeling of human-like emotions and human behaviors. We have started to build a piece of future.
When you add emotions to pseudo-human intermediaries between the user and the software application, you must also consider how different cultures express and handle emotions. What is your approach to this challenge?
This is easy: user profiling.
Interacting with a software application or database at a level so close to natural language requires trust in your technology. For example, if I check my bank account using one of your human-like interfaces, I want to be sure my questions are fully understood and the answers I receive are accurate. How do you build trust in your users?
From our experience, trust results from the level of the perceived professionalism of the technology a user is confronted with and the familiarity of the interface he’s using to interact. We build on professionalism by using AI to enable our K-Humans™ virtual assistants disambiguate and identify users’ real needs. We are the layer in between humans and machines: we therefore interact with users to ensure that the questions they ask correspond to what they really desire, and help them complete their request with the data they eventually didn’t think of or remember. By enhancing all the human aspects of our K-Humans™ virtual assistants, such as displaying emotions or having a sense of humor, we build immediate confidence and familiarity since users find in them the behaviors of real persons. Our users always refer to our assistants calling them by their name: this says a lot about their way of approaching our technology.
The integration between the knowledge base and the interaction technologies (avatar and similar) position your company among the most advanced in this area. Do you think promoting innovation from Italy is more difficult than it would be promoting it from here in the Silicon Valley?
I believe every market has its own prerogatives. Silicon Valley sure is an enriching place, it would be interesting to see the inevitable synergies that would derive from being close to so many innovative companies.
You are considering creating an outpost of your company here in California. What are your plans and expectations for this idea?
We think that the opportunity to develop virtual humans is just at its beginning. It’s obvious that the US is a seductive market, but what is really relevant for us is to identify who will lead in the next steps of this evolution. Our K-Human™ virtual assistants are in pole position right now, but to stay there we need to cooperate with the best minds, companies, and universities on the planet. We perceive this field as something that is accelerating more every single day. Today the market is focused on specific applications, such as call center applications, info-points, or educational tools; but in a few years who knows what our experience could be used for: the movie industry, video-games, or -why not- developing a physical human-like virtual assistant.
I would like to thank Giorgio Manfredi for taking the time to speak with me today. If you have any questions for Giorgio or for BAIA, please leave a comment below and we will be glad to answer.
Franco Folini 