Tuesday, March 12, 2019
AI is reshaping the world. Is reshaping the way we interact with one another and inevitably the way we live. AI is not something new as it was introduced as an academic discipline back in the late 50’s. Nowadays revisited, it has given life to new “intelligent” applications that usually live in the cloud, are accessible through smartphones and tablets and offer us a whole new experience in the way we are self-served and do business. Bots!
One important feature of bots is their user interface which allows a person to interact with them using natural language either spoken or written. Another one is that they are ubiquitous, ready to assist and serve any time of the day through almost any channel. The epitome of features though is that they know and remember things… Machine learning and knowledge base capabilities give them "wisdom" and sometimes make them more efficient than humans.
Bots are ideal when you need to order something through an itemized and well-defined inventory but are lame when you must use more abstract concepts. They are good for structured dialogs and evaluations but still inappropriate for completely free speech where anything can be said or typed. This happens because the technology is still at its infancy, hence the results are better and the bots are more efficient when the dialogs and concepts are domain specific, like for instance banking or telecommunications customer service.
Bots for specific domains are easier to design and implement but what one must always have in mind, is that implementing a conversational self-service application is not just another IT project. Most importantly it is a digital transformation project as it will change the way an organization will approach and interact with customers. In that sense there are a few things an organization must pay attention to when implementing a bot.
The dialogs must be carefully designed to offer a human-like conversational interface. The dialog at any point must avoid dead-locks like when a human is talking to another human.
The bot must be context aware and “remember” what was previously discussed. Moreover, it must be able to understand what the customer is talking about, even when the request becomes somewhat more abstract and the customer implies it.
Since a deadlock in the dialog will lead to bad experience, escalation to a human must always be an option, even if this means the interaction will continue through a different channel.
People want to be serviced fast and accurately, so a bot must be able to provide answers by being able to look-up information in a KB using as search keys the dialog context and concepts captured within it.
People like to feel important and unique, meaning that the next time they will have to chat with the same bot, it must somehow remember them along with their preferences and habits, like a real person would remember.
The bot is nothing else but an application which captures and generates a lot of data. These data can provide important information and valuable insights to organizations if properly analyzed.
They can indicate design flaws and show the correct way to fine-tuning and evolution of the bot.
They can show the path the customer followed to reach his goal and whether that was even or a bumpy one.
They can provide hints for new services and things an organization needs to improve in its products and services.
The transformation has obviously internal aspects as well. Given the fact that the bots will replace humans in simple tasks, means that the organization processes will transform as well, and most probably the organization will meet internal resistance.
As consumers, we will meet and interact with a bot when we will call an organization’s contact center or visit its web site for assistance or information. Bots will reach out to us to send products and services promotional messages through various channels and will become our Facebook “friends”. Today, most of us will gladly accept a “friendship” request by our favorite shoes or technology firm as we would like to be aware of new products or special offers. A few years ago, we might never allow such a thing to avoid spamming. Thanks to machine learning, the intelligence of the bots makes contact less intrusive as they learn our preferences and habits and the frequency and kind of contact we would like to have.
Marketing and sales will also benefit from bot evolution but it’s more than that. Think of a bot that will interact with your organization’s personnel to answer questions related to compensations and benefits, give approvals to access systems or fulfill certain requests. How much money could save from your HR a bot like this? Think of an insurance consulting bot that will assist customers find their way around with claims or a financial investment consulting bot that will help you with your next investment.
Possibilities and applications are limitless, and this journey has just begun. Bots are here to stay and make our life easier. The tech behind them is evolving fast and is making them smarter. They will soon become an essential component of every IT solution, not only because they will be able to do things faster and more efficiently but also because the economies expected to be achieved are predicted to be vast. Expensive to build? Yes, but the ROI will be fast and the realized non-financial collateral benefits enormous.
Say hello to our new “friends”, say hello to Bots!
Author: Thanasis Mercouris.