By Greg Lush
"Use your words" was often said to me growing up as a young child. Now, it is not what you think — I was a relatively good kid. My identical twin and I formed a unique language that only he and I could understand. For those moments when my brother and I were not bickering, we would communicate using these words, which compelled our parents to remind us to "use our English words.” Our older sister was pretty good at interpreting, she would correlate the bundling of letters, often forming words, to our actions. Essentially, she was understanding our utterances and creating intent, not too different than the natural language processing engines we are just now starting to enjoy as part of our digital landscapes.
If you or your team are the ones tapped to introduce BOTs and Cognitive Speech services to your organization, welcome to a fantastic ride. Assuming you have been following along my prior posts, you should be good to go. A stronger engine does not exist than BOTs to wrap all your enterprise data into a nice package which is accessed by simply asking a question. Do you remember years ago when you waited for the newspaper or the morning news to understand what the weather would be like today? Now what do most of us do? Ask our digital assistants, "Hey XYZ, what does the weather look like today?" Did you feel a bit awkward asking a computer a question? It is interesting when I visit with folks, they seem to ask simple (and limited) questions to computers, almost like we are not very sure how the computer would like us to form the question to seek the proper response. Often we throw in words that we think the BOT will understand forming sentences that we would never verbalize to another human being. For crying out loud, stop the madness! With BOTs, and especially those more sophisticated approaches integrating natural language processing, you should be comfortable asking it a question just as you would a colleague. How the BOT is configured is critical, and your automation targets should be clearly defined and as narrow a scope as possible. Consumers of this technology are not forgiving, they will try once and if the return is not relevant, will often discount the value and may never ask the BOT a question again. Understanding that the sentence seems a bit dramatic, make sure that you have meaningful content which has been trained, repeatedly. You will only have one chance to impress! Three approaches are provided that will help you get started, they are organized by level of complexity.
1. Questions and Answers
Who has not heard of a frequently asked questions (FAQ) list or forum? Interestingly, back in the 70s a book was written suggesting the power of contextual, per software screen assistance known as Electronic Performance Support Systems (EPSS). Crazy to think that just in recent years the FAQ or Help screens have evolved past functional, and often disconnected from business need, contextual references and guidance. Although many computer users are disenchanted with Help, and they may refuse to return, this is a great opportunity for the simplest level of computer automation with a BOT. Take those FAQs and Help screen contents and work them into a question/answer-based BOT sequence. Add a bit more sophistication by adding natural language processing so you can encourage users to get comfortable asking questions as they would to a trusted colleague.
Delivery of the BOT content can vary widely between text messaging, web pages, embedded within Intranets, etc. For me, if I am asking questions regarding the use of a computer it is always easier receiving that information on another digital tool. Texting responses, with specific instructions or details, seems to work very well. Recipients may then use their phones to ask questions and the computer/tablet to run through the steps or advice. Most BOT engines will connect to an inexpensive tool like Twilio or other SMS service, providing dedicated numbers in a medium that most are comfortable, texting.
2. Business Process
"Let's not boil the ocean" is an idiom often said when enthusiasm meets the pragmatist. Those close to me would say I lean closer to the enthusiasm or dreamer role; however, do not hesitate to refer to me as a pragmatic technocrat. Much of my career, after I came in from the field, was filled with business process assessments. Heck, you need a few different skills to help an organization evolve into a digital enterprise. A passion and curiosity of available technology tools; listening and question asking skills, the ability to see the big picture, and business process mapping. Once high-level business objectives and strategies are mapped it is important to break these down to micro-process improvements. No business, at least that I have ever seen, is not in need of some type of business process engineering. I suppose two books, which you may find interesting, have influenced my perspectives for years, they are:
- The Goal [Eli Goldratt]: a book read by many in manufacturing; however, the concepts suggested applying to every business. Eli discusses and provides examples of the "theory of constraints," suggesting that you focus on and resolve one constraint, a new constraint will appear. A never-ending process of continuous improvement.
- How to measure anything [Douglas Hubbard]: fantastic book for anyone and reminds us that anything can be measured.
Deploying process modification within a business without clear objectives and points of measurement is a fool’s errand. When it comes to BOTs and process automation this could not be more important. Sure, many things within a business could stand shoring up; however, which will have the greatest value on your investment? In a perfect world your first process automation target will fund future BOT and process re-engineering efforts. As a friend and former co-worker used to remind us; make informed and deliberate decisions. Prioritizing the business value and potentially choosing an area of improvement which is not that "sexy" is a reality you should be willing to accept in order to be successful. If you only hunt elephants, you may starve while tripping over the rabbits.
3. Industrial IoT
For years my work has been focused on elevating the conditions for the last mile of the field workers. One of my favorite areas of digital interest has been around taking advantage of the consumerization of equipment-based digital fingerprints. Many refer to this as IoT or the Internet of things. While the concept of IoT is not new the amount of information which we are collecting, essentially based on the affordability of the event producer, is tremendous. However, we now must make this information actionable in a meaningful and relevant way which compels a different outcome at the point of service. This interaction with the field worker is the area which has piqued my interest regarding the use of conversational bots.
Sometimes people confuse these bots with sophisticated search algorithms, this is not the area that had any interest to me. Instead, the bot, in my mind, can be used to take advantage of and make meaningful the correlation between a multitude of transactional systems (work order, accounting, CRM, etc.).
Transactional systems can reside either within or outside of your computing environments. One of the greatest advantages, which struck me immediately, was the ability to simply carry on a conversation with a field worker. In the background predicting and guiding the field worker through bots-based responses. Why should a field worker be concerned with what system they must enter this… or go to find that?
As you begin to explore the technology which offers this humanistic conversation-based approach, you will also realize that our possibilities are endless. Not only can we apply these bots, and their underlying algorithms and machine learning routines to field workers, but also apply it to sales, project crews, etc. Anyplace that we need to correlate and leverage transactional systems in our daily working routines. So, are the bots coming? Yes, I believe in the next few years we will see bots everywhere performing small functions and helping us consume multiple perspectives; so, we as humans can make better decisions. Are you ready?