WIXLIB

Conversational Chatbots – Let's chatJune 2017

ConversationalBrochure/ FinanceChatbotsin the-DigitalLet's chatage Section title goes here 02

Conversational Chatbots - Let's chatAgendaIntroduction 05Do Chatbots have inherent advantages overhumans 07Few considerations which enterprises may look atbefore evaluating chatbots 08Identifying chatbot use cases 09Why are chatbots steadily gaining attention? 10Enterprises are finding multiple areas to deploy chatbots 1103

Conversational Chatbots - Let's chat04

Conversational Chatbots - Let's chatIntroductionChatbots are computer programs thatmaintain a conversation with a user innatural language, understand the user’sintent and send responses based on theorganization’s business rules and data.These chatbots use Artificial Intelligenceto process language, enabling them tounderstand human speech. They candecipher verbal or written questionsand provide responses with appropriateinformation or direction. Many customersfirst experienced chatbots throughdialogue boxes on company websites.Chatbots also interact verbally withconsumers, such as Siri on the AppleiPhone or Amazon’s Alexa Voice Service.Chatbots are now increasingly being usedby businesses to augment their customerservice.Building blocks of chatbots:The chatbot consists of these keycomponents: A front-end interface, which connects toa variety of channels, such as websites,email, SMS, or messaging applicationssuch as Facebook Messenger or Slack,through which users interact with thechatbot. Understanding intent is responsiblefor recognizing the user’s intent.This element uses natural languageprocessing and machine learning toparse user messages, collect relevantparameters from words and sentences,and map those to actions to take. Another component managesthe dialogue by maintaining arepresentation of the conversationallogic and keeping track of context.Chatbots can be considered as anenhanced channel of customer interactionwhich would move from InteractiveVoice response to Intelligent AssistantResponse.So let’s take a look at the evolution of chatbotsEvolution of a chatbotBasic Scripted Simplistic chatbot Looks for key phrases &give scripted responsesExampleRequest: Show me a list ofall term insurance policies.Response: Bot lists outterm life insurance planswith options to make aselection. Bot capturesuser's selection in a form,and presents next optionsas per Q&A knowledgebase.Source: Deloitte AnalysisIntent Recognizer These systems try toidentify parts of speech foreach word and how thewords relate to each otherto extract meaning from arequestExampleRequest: Show me cheapterm plans?Response: Bot will ask theuser the coverage amountin order to compare thepremiums befor listing outthe plans.Dialogue ManagerIntelligent AdvisorHuman Advisor Maintains past conversations Ability to understand what a Gerneal AI Can integrate gesturehuman is trying to achieve.(mythical stage)understanding with language Could advise on betterunderstandingpaths.ExampleRequest: I want a coverageof 1M.Response: Bot lists out termlife insurance planswith premiums.Request: I don't want tospend more than 500 peryear.Response: Bot understandsthat the user is talking aboutlife insurance premiums andupdates the list based onpremium limit specified by theuser.Example:Response: Prepares acomparison table of affordablelife term insurance plans alongwith the payment options andbrochures to download.In addition, the bot suggests thatthe user should consider morecoverage.It is advisable to have life insurance atleast till your target retirement date.Do you want to see premiums for 20year policies?05

Conversational Chatbots - Let's chatThere are two kinds of chatbots in themarket today: Scripted bots and ArtificialIntelligence (AI) bots.Scripted bot is like a rule-based guidedconversation and performs like a decisiontree where each action by the userprompts the bot to take action or respond.For example, Sephora (a popular makeupretailer in the U.S.) has a bot on Kik. The06bot engages users with questions aboutmakeup preferences and serves up contentand offers relevant to their responses.While it may not sound like a sophisticatedprocess – the more the consumer engageswith the bot over time, the smarter thebot (and the brand) becomes aboutconsumer preferences and is able to servepersonalized content and offers.AI bots are built on Machine Learning(ML) and Natural Language Processing(NLP) capabilities. They are based on thehuman capability of learning and absorbinginformation owing to which they are moreefficient and can process much fasterthan humans and may come up with moresubtle results.i

Conversational Chatbots - Let's chatDo Chatbots have inherentadvantages over humansChatbots help enterprises in variousways as mentioned below:1. Chatbots have a number of potentialbenefits over traditional GUIs. First,they can simplify applications for users.For example, rather than navigatingthrough an interface or website tofind information, users can just say ortype what they want. Users can alsocompress multistep tasks into a singlecommand, such as, “Get my list of openopportunities this quarter, and send itto Janet.”2. Second, the conversational UIs thatchatbots offer may require little to notraining, given that they understandand can interpret natural language andtranslate it into actions.3. Third, users can leverage chatbots tooperate several business applicationsat once. For example, users can invokemultiple chatbot actions in conversationwith team members at the same time.Combined, those benefits allow for nonexpert users to interact with many complexapplications in an intuitive fashion ina single interface. This gives rise topowerful automation opportunities,where chatbots trigger actions andorchestrate processes across a rangeof applications through the courseof dialogue in natural language. Thebusiness impacts can include reducingcosts by increasing self-service, improvingend-user experience and satisfaction,delivering relevant information faster,and increasing compliance with internalprocedures.ii07

Conversational Chatbots - Let's chatFew considerations whichenterprises may look atbefore evaluating chatbotsAs the adoption of voice and messagingtechnologies continue to grow in theenterprise, so does the potential forchatbot solutions. However, in order tobe adopted in the enterprise, chatbotplatforms will need to provide capabilitiesin areas such as integration, securitymanagement, or monitoring, which areessential elements of enterprise solutions. hether the identified businessWprocess can do away with humaninteraction? - It is important tounderstand that although AI and NLPbased chatbots are getting intelligent,there is still scope for improvement.E.g. Chatbots when initially configuredhave a limited dictionary based on thetraining examples provided. As the botis used more and more it encountersdifferent live business situations whichmay be different from the trainingexamples, the bot learns and improvesits accuracy. I ntegration with enterprise systems/ CRM tools – Can the chatbot toolintegrate with existing enterprisesystems? To streamline operations,enterprise chatbot platforms shouldprovide a simple way for chatbotdevelopers to integrate with existingbusiness systems and translate the datainto voice or simple text interfaces. overnance or monitoring systemsG- An enterprise-ready chatbot platformshould include monitoring capabilitiesto track the different command andresponses of its users and consumerapplications. Security - To effectively leverageenterprise data sources, chatbot08platforms should provide securitycapabilities such as encryption, dataprivacy, and access control policies. irtual Personality - AlthoughVcustomers often realize they are talkingto a machine when using a chatbot,many people are more comfortablewith a virtual human experience suchas an avatar. Customers appreciateexpressions of empathy and patience.You can program your chatbot withpersonality that reflects your company’sbrand, style and values. andover to Human - CustomersHoften become frustrated when theycan’t access a human for customerservice. To address this, maybeenterprises can add a button in thechatbot text window that a customercan click on, or provide a direct phoneline, to enable them to chat with ahuman customer service representativeat any time.08

Conversational Chatbots - Let's chatIdentifying chatbot use casesThe key factors to assess suitability of using conversational UI are:Repetitive procedures - Professionals frequentlyand repeatedly perform a task or exchange the sameinformationOnerous current processes– It is time-consuming,difficult, or confusing to navigate current processesRequires human interaction – Current processesrequire engaging expensive resources whose timecould be spent on higher value activitiesEasy to consume and control via conversation–Workflow tasks or information exchanges are suited tothe back and forth nature of natural conversationThe key factors to assess viability of using conversational UI are:09Complexity of Rules / Standardization- Businessrules are well-defined and processes are standardizedTechnology Compatibility- ease of integration withcompany systems or databases via existing APIsData Availability – Transcripts of human dialogueare available for pattern recognition or a database ofcommon questions and their answers is availableFirm Readiness- meeting requirements of security,accuracy, ease of use, price, scalability, etc.09

Conversational Chatbots - Let's chatWhy are chatbots steadilygaining attention?Advancements in technologies andcomputing power have enabled chatbotsto gain traction from major technologygiants as early adopters. With advances inNatural Language Processing and ArtificialIntelligence, machine learning algorithmscan now understand queries just like thehumans. More importantly the code learnsfrom every interaction and grows its skillsto handle further conversations.Let’s look at an example of how HumanResource function can benefit from AIchatbots. Human Resource companiesinformally refer to Chatbots as a friendwho helps you with tasks such as bookinga flight ticket, reminding you to punch infor your attendance, sending a note toyour newly joined team member or eventraining you for the next job interview.Some chatbots can help with employeeperformance reviews and engagementwhile another can simplify the recruitmentprocess through Artificial Intelligence.iiigaining attention. Some of these trendswere: Mobile messenger domination Mobile messengers such as FacebookMessenger, WhatsApp, WeChat andother have become the preferredmeans of communication on mobiledevices. Chatting has emerged as thecommunication medium of choicefor today’s smartphone carryinggeneration. App fatigue - App fatigue refers toconsumers who are tired of installing,updating and learning how to usemobile (iOS and Android) apps, and tobusinesses that are tired of spendingheavily on mobile app development,maintenance and marketing only to findout that very few people actually usethem. Push by global technology majorslike Facebook and Microsoft Microsoft announced its Bot Frameworkand Facebook announced their supportfor bots in the Facebook Messenger.–– Chatbots integrated intomessengers are carrying targetedcustomer communication in theform of text messages. Popularmessaging platforms are comingup with APIs which allow chatbotdevelopment and integration withintheir messenger apps possible.Most Chatbots use Machine Learning tolearn from the initial interactions with theemployees. The first time that a chatbotencounters a question which it cannotanswer, it is programmed to pass thesame to the next level which is a humaninterface. This response is recorded by thebot for future reference.As shown in the evolution chatbotshave existed since 1960’s without goingmainstream. Few recent global trendshave made 2016 the year chatbots started1010