<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by MindTitan on Medium]]></title>
        <description><![CDATA[Stories by MindTitan on Medium]]></description>
        <link>https://medium.com/@mindtitan-ai?source=rss-69684ed99bb2------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/0*2zuJEBHJSIgiruij.jpg</url>
            <title>Stories by MindTitan on Medium</title>
            <link>https://medium.com/@mindtitan-ai?source=rss-69684ed99bb2------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Tue, 17 Nov 2026 05:11:14 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@mindtitan-ai/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Call center automation with AI — how to actually make it work?]]></title>
            <link>https://mindtitan-ai.medium.com/call-center-automation-with-ai-how-to-actually-make-it-work-65db62de5f2d?source=rss-69684ed99bb2------2</link>
            <guid isPermaLink="false">https://medium.com/p/65db62de5f2d</guid>
            <category><![CDATA[customer-experience]]></category>
            <category><![CDATA[call-center-automation]]></category>
            <category><![CDATA[customer-service]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[machine-learning]]></category>
            <dc:creator><![CDATA[MindTitan]]></dc:creator>
            <pubDate>Wed, 28 Jul 2021 15:54:36 GMT</pubDate>
            <atom:updated>2021-08-19T19:14:44.090Z</atom:updated>
            <content:encoded><![CDATA[<h3>Call center automation with AI — how to actually make it work?</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*wjV_GpeLzpWmaa2I.jpg" /></figure><p>Managing a call center is a difficult feat. If you haven’t been in a contact center or call center, picture this to put it into perspective: there’s an office filled with hundreds of agents and managers, fielding tens of thousands of calls, emails, and chats from thousands of customers. Taking it further, the agents are often new to customer service, have little to no professional training, and their average tenure is fourteen months. The agents are under pressure to solve issues or requests quickly while handling angry or upset customers. To add to the challenge, customer service departments strive to give every customer <em>the utmost experience possible</em>.</p><p>This article explains contact center automation, why it is important to start thinking about automation, and what are the different automation alternatives. We go through a step-by-step process of call center automation with artificial intelligence and demonstrate real use cases with examples from the Telecommunication industry.</p><h3>What is call center automation?</h3><blockquote>Call center automation is the process of fulfilling repetitive tasks with the help of technology.</blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*TLRMahlS1HGhFHJm49AhxQ.jpeg" /></figure><p>This can be full automation of calls or semi-automation by increasing the efficiency of human agents. The most advanced call centers use AI-powered solutions with natural language processing (NLP), natural language understanding (NLU), and speech recognition models for full<a href="https://mindtitan.com/products/call-automation/"> call automation</a>.</p><h3>Why is call center automation important?</h3><p>Call center automation has become especially critical during the Covid crisis. According to CGS research,<a href="https://www.cgsinc.com/en/resources/infographic-ongoing-impact-covid-19-contact-center-support-services"> 74% of CX leaders reported</a> a significant increase in inquiries in their contact centers during the pandemic. This brought about longer waiting times, higher call abandonment rates, and lower first contact resolution rates (FCR).</p><p>Even though a sudden increase in call volumes has a temporary effect, it actually can happen to any organization at any time simply due to a new service, a policy change, a technical failure, or even a merger of two organizations. No one can predict the occurrence of all such situations, but undoubtedly businesses should be ready for a sudden increase in customer inquiries or planned scaling of the business.</p><h4>1. High call abandonment rate</h4><p>A high call abandonment rate can cost companies hundreds of thousands of euros annually, as some of the clients that try to reach the call center might want to buy something. For instance, based on our experience with Telecom, a 20% abandoned call rate actually costs 2% of potential sales opportunities, as on average 10% of the calls that were abandoned were related to sales. Of course, this number can vary from niche to niche, but call abandonment is definitely a problem that requires attention.</p><h4>2. Bad customer experience</h4><p>From the client’s perspective, long waiting times can negatively affect customer experience. No one wants to wait several minutes on the line until they are connected to a virtual agent and a bad experience definitely leads to higher customer churn rates.</p><h4>3. Cost optimization</h4><p>Cost optimization is another argument for automated customer interactions. The cost of contact in developed countries can vary from EUR 3 to 10 per contact and sometimes even more. Just imagine the effect on a contact center handling 100,000 calls per month spends EUR 300,000. So what could be a potential cost optimization if AI could automate 20% of the calls? We are speaking about hundreds of thousands per year.</p><h3>Call center automation alternatives</h3><p>There are several alternative automation processes that can be integrated into a contact center. In theory all of these decrease the typical contact center problem of long waiting time.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*PpcAh6X-829FuUM_wfBmFg.jpeg" /></figure><ol><li>Self-service would allow customers to find the necessary information on their own without actually contacting a call center. The problem is that on average only<a href="https://www.gartner.com/smarterwithgartner/rethink-customer-service-strategy-drive-self-service"> 9% of customers</a> that use self-service actually resolve their issues, again leading to a decrease in customer satisfaction and an increase in the number of inquiries.</li><li>AI-powered <a href="https://mindtitan.com/resources/guides/chatbot/">chatbot</a> that could potentially resolve most of the customer inquiries. In fact, based on the example of one of the biggest Telcos in Nordics,<a href="https://mindtitan.com/resources/case-studies/elisa-case-study/"> Elisa</a>, a chatbot can actually handle 60% of all written inquiries, out of which half can be fully resolved without any agent assistant. However, a chatbot cannot be an ideal solution and alternative to a contact center, simply because <a href="https://edition.cnn.com/2011/09/22/tech/mobile/americans-prefer-text-messages/index.html">53% of consumers</a> still prefer to call, instead of texting a virtual agent or a chatbot. The manager of a contact center at a prominent Northern European bank noted that they did not experience an <em>immediate</em> decline in customer contacts after they introduced live chat to their website. Hence, the volume of customer contacts increased after they deployed online chat, while the number of customer service agents remained the same.</li><li>Simplifying the IVR. As customers, we don’t like too many options when we reach a contact center, and customer service managers understand that, so they simplify. There are usually three, but there can be up to about ten options. The goal is to get the customer to the correct agent as fast as possible. However, based on the work MindTitan has done with contact centers, on average customers wait 2.4 minutes on the line before they are connected to the right agent. Whether this happens due to a complicated IVR structure (as in fact, in 30% cases customers press the wrong button) or simply because of a misclick by a customer, it all leads to poor customer satisfaction and a high abandoned call rate.</li><li>Hiring more people. Well, some organizations do believe that opening a new contact center and hiring additional people might change the status quo. However, there are some problems you should consider before you choose this option.</li></ol><p>First of all, pay attention to the agent turnover factor, which is considered the number one challenge for contact centers. As agents should often know answers to hundreds of different questions, the cost of staff training is high.</p><p>Secondly, managers do not have proper tools to monitor the quality of the actual conversations with clients, as they only have the capacity to choose some random calls to listen to. When you have several hundred agents in a contact center, quality monitoring becomes <em>mission impossible</em>.</p><p>Both of the mentioned factors lead to a situation where it is difficult to sustain the service level agreement (SLA). Particularly because the new staff is constantly coming in and on average it takes 6 months of working experience and training to be well prepared for actually solving customers’ issues. Considering that the average agent turnover is fourteen months, this becomes a real challenge. We have<a href="https://mindtitan.com/resources/blog/customer-service-success-revealed-by-elisa-challenges/"> interviewed Mailiis Ploomann</a>, Head of Telecom services at Elisa, who mentioned the optimal number of contact center agents for customer service that is possible to manage efficiently without a decrease in quality of service.</p><h3>Steps for contact center automation with AI</h3><p>Contact center intelligent automation can be achieved in several steps. The exact number depends on the complexity of the customer service and contact center operations. Let’s go through the basic ones.</p><h3>Step 1. Get rid of the traditional IVR</h3><p>Wouldn’t it be much simpler, if a customer would just explain the reason for the call with their own words without the need of pressing any buttons? Often people are confused by the complicated structure of the old legacy IVR, that it is actually not very efficient anymore. <a href="https://mindtitan.com/products/voice-ivr/">AI voice IVR</a> simply asks the customer, “What can I help you with?” as soon as the customer has reached the contact center.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*uqNWWgMUoZB1w45wjZ5_rQ.jpeg" /></figure><p>As the customer describes the issue in their own words, the machine learning model listens and transcribes the call in real-time. Since the model is taught using previously recorded data, it knows which words and phrases are associated with specific customer issues — or reasons they’re calling. Based on the classification, the call is routed to the right agent the first time. And the model only gets better as it categorizes call event data, which is checked by a<a href="https://mindtitan.com/resources/blog/human-in-the-loop-machine-learning/"> human in the loop of machine learning.</a></p><p>This significantly helps to improve customer experience as it decreases wait times and makes it possible for the business to prioritize particular clients, e.g. sales opportunities, customer retention, or technical failures.</p><h3>Step 2. Increase efficiency of your agents</h3><p>Call centers face problems of agent retention and costs of new staff training. As a business is trying to provide a better customer experience and fulfill the SLA’s, it needs to have a proper tool to support its agents. For example, the<a href="https://mindtitan.com/products/cs-assist-tool/"> CS assist tool</a> helps agents to ask the right clarifying questions and give the right answers by following the conversation in real-time using enhanced customer-specific AI models.</p><p>CS Assist tool works in combination with AI Voice IVR, so when a client is connected to customer support, the agent already sees the transcription of the call and the most probable reasons for the contact classified by AI. The agent should simply follow the solution flow and read out the text. Check out the video below that demonstrates that in action.</p><p>For solving inquiries, AI uses the knowledge base called solution flows that allow making all kinds of different API integrations with the backend systems. You only need to keep an eye on the first contact resolution rate (FCR) and make solution flow improvements until you get optimal numbers, as otherwise, these will not actually solve the customer’s problem. From our experience, it takes around 4 attempts to reach an optimal FCR. What is even better, the same solution flows can be shared with the chatbot and all other customer service channels, making it a unified knowledge base for solving customer issues.</p><p>Implementation of the CS Assist tool helps to significantly decrease the onboarding time of your staff, as agents will simply follow the solution flows and read the text. The tool helps to meet SLAs, and decrease the average handling time.</p><h3>Step 3 Automate the call</h3><p>As soon as the previous 2 steps are fulfilled, you might be thinking about<a href="https://mindtitan.com/products/call-automation/"> fully automating customer interaction</a> over the call. When the solution flow reaches a good level of FCR, it becomes possible to integrate a speech synthesis model for full-contact resolution without any human assistance.</p><p>To make it happen, you need to fulfill the main condition:</p><ul><li>Solution flow should achieve a high level of FCR. From our experience, it is around 80%.</li></ul><p>The CS assist tool allows you to build solution flows for various questions, so as soon as some of them are tested and have reached the FCR of 80, they become good candidates for intelligent automation. As soon as this is achieved and previous steps fulfilled, you can start fully automating customer interaction with the contact center.</p><h3>Step 4 Get customers forwarded to the chatbot with an SMS</h3><p>Wait for a second, we just stated that it is possible to automate contact center interactions with AI, why can’t we automate all inquiries as described in the previous step? Well, we have an answer to that question. A phone call has technical limitations and compared to the chatbot you have an additional step which is the call transcription model that turns spoken language into writing. In a nutshell, this means that you can automate fewer questions over the call than by using a chatbot.</p><p>So how does it work? Well, that is very simple. When the customer calls and explains their issue, first the AI model classifies the topic of the question as mentioned in previous steps. If at this moment AI detects a possibility to fully resolve the issue over the chatbot (whereas solving this over the phone call is not possible or will be time consuming), AI will ask the customer whether it is possible to send the link to a chatbot with an SMS that will solve the issue. If a customer agrees, they will automatically get the link to a chatbot. If not, the call will be transferred to the right agent and solved using the CS assist tool.</p><p>This additional feature helps to increase the percentage of automation and gives an opportunity for customers to interact with a chatbot, experience a good quality of service, and probably use it in the future as a channel for communication.</p><h3>Conclusion</h3><p>This article outlines the importance of call center automation and lists different methods that can be integrated for better customer interaction with the contact center. As discussed, call center automation is the process of automating some repetitive tasks with technology. One of the purposes of automation with AI is to give an agent a possibility to focus on the most important things, where they are most needed, and make their work more efficient.</p><p>Businesses should also take into account the fact that automation saves a lot on costs, as today an average cost per contact is around EUR 3. Automation also helps to decrease the abandoned call rate to nearly zero, whereas a 20% call abandonment rate costs companies 2% of potential sales opportunities.</p><p>While a single machine learning model improves the customer experience, combining multiple models across customer service and bringing in data from outside the contact center brings true efficiencies while vastly improving the customer experience. We recommend reading our comprehensive guide that outlines how customer service can be<a href="https://mindtitan.com/resources/blog/automated-customer-service/"> truly automated</a>.</p><p><em>Originally published at </em><a href="https://mindtitan.com/resources/blog/call-center-automation/"><em>https://mindtitan.com</em></a><em> on July 28, 2021.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=65db62de5f2d" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Artificial Intelligence for Government: How Estonia is Leading the Way]]></title>
            <link>https://mindtitan-ai.medium.com/artificial-intelligence-for-government-how-estonia-is-leading-the-way-a413860d0666?source=rss-69684ed99bb2------2</link>
            <guid isPermaLink="false">https://medium.com/p/a413860d0666</guid>
            <category><![CDATA[public-sector-innovation]]></category>
            <category><![CDATA[use-cases]]></category>
            <category><![CDATA[ai-in-government]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[ai-for-public-sector]]></category>
            <dc:creator><![CDATA[MindTitan]]></dc:creator>
            <pubDate>Wed, 14 Jul 2021 12:54:26 GMT</pubDate>
            <atom:updated>2021-07-14T12:54:26.490Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*292BA3OWsAQ3CkuSzWvUlA.jpeg" /></figure><p>It’s hard to believe that just 30 years ago, the small Baltic state of Estonia was lacking many of the services modern governments take for granted. Nowadays, the country has not only bounced back from decades of Soviet administration but become a model for the way digital transformation can boost efficiency, security, transparency and inclusivity in the public sector. What’s more, innovation is booming: Estonia boasts <a href="https://e-estonia.com/lessons-from-the-worlds-startup-capital/">six times more start-ups</a> per capita than elsewhere in the European Union.</p><h3>Artificial intelligence for government include:</h3><ul><li>Automated communication with citizens</li><li>Tax fraud reduction</li><li>Personalised learning path in education</li><li>Prediction in road traffic accidents</li><li>Emergency response and assistance</li><li>Tracking illegal ads in social media</li></ul><p>99% of public services in Estonia are available as e-services, and Estonia continues to invest heavily in technology and <a href="https://e-estonia.com/estonia-accelerates-artificial-intelligence/">artificial intelligence (AI)</a> for the public sector. Let’s take a look at some of the most interesting artificial intelligence government applications currently being explored in Estonia.</p><h3>Artificial intelligence for government can streamline communication</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*hRU9HRWaMX4yfaeCrzxXsA.jpeg" /></figure><p>Named “Kratt” after a creature from old Estonian folklore, the 2019 Estonian AI strategy aims to boost the uptake of AI by both the public and the private sector. The strategy will invest at least 10 million euros over a three-year period.</p><p>The Estonian government has laid out a vision for how it intends to use AI to benefit both the public and private sectors. As part of the strategy, the government will take a leading role in promoting and accelerating the use of AI-based applications. #KrattAI will focus on four main pillars, boosting the uptake of AI in government, the economy, research and development and the legal sector.</p><p>In order to develop the strategy, the government looked at how citizens interact with public services and the challenges they can face. It found that people are often uncertain of which government agencies they need to contact. What’s more, they tend to be unaware of many opportunities offered by the government. Communication between public service agencies and citizens was more often than not time-consuming and ineffective.</p><p>In order to make public services easier to use and to better structure communication, #KrattAI aims to create a system in which citizens will be able to get all the information they need in one session, using one device, via a virtual assistant. The strategy foresees the development of open-source chatbots, an authentication system that will enable the use of private sector digital channels (such as WhatsApp or Facebook Messenger), speech synthesis applications and a consent management platform.</p><h3>Reducing tax fraud with artificial intelligence for government</h3><p>Taxes enable governments to finance public sector initiatives and create national environments that are conducive to economic growth. Unfortunately, tax fraud remains a major issue in Europe. According to official statistics, European Union member states lose up to <a href="https://www.europarl.europa.eu/doceo/document/E-9-2020-004736_EN.html">190 billion euros</a> a year to tax avoidance and evasion. Estonia is no exception: the Estonian Tax and Customs Board (ETCB) calculated that in 2019, the nation had lost 134,1 million euros of tax revenue in non-reported or under-reported income.</p><p>Unfortunately, the processes currently employed by many national tax agencies, such as document processing and reporting, are time-consuming, repetitive and slow. Artificial intelligence can help agencies detect fraud more efficiently with fewer false positives, enabling public servants to dedicate their time to high-value tasks while building a trust-based relationship with taxpayers. The ETCB is currently seeking to create an <a href="https://mindtitan.com/resources/blog/mindtitan-develops-an-artificial-intelligence-model-for-the-estonian-tax-and-customs-board-to-reduce-tax-fraud/">AI-based “human-in-the-loop” system</a> that will not only enable more efficient tax collection but also increase accuracy over time.</p><p>This will enable more accurate identification of potential fraud and streamline operations of both analysts and auditors.</p><p>According to the OECD, AI will eventually free up nearly <a href="https://www.oecd-ilibrary.org/governance/hello-world_726fd39d-en">one-third</a> of government agents’ time. Indeed, one of the major benefits of AI is its ability to enable tasks to be performed more efficiently by fewer people. This is of particular interest to law and tax enforcement authorities, which are often faced with high volumes of appeals to process and limited human resources.</p><h3>AI can help create personalized learning paths for students</h3><p>Education is another field in which resources are limited, but needs are consequential. Each pupil is different, yet teaching in public school systems the world over tends to be geared to the needs of the “average” student, rather than customized to individual learning requirements.</p><p>The Estonian Education and Youth Authority has begun to explore the use of AI to create <a href="https://www.hitsa.ee/personalized-learning">personalized learning infrastructure</a>. The infrastructure aims to support the creation of customized learning paths without placing an extra burden on the teaching staff. To do this, learning analytics are used to interpret students’ learning behaviour, i.e the way they learn, what inspires them and what holds them back. The infrastructure will provide students with personalized suggestions to improve the way they learn, and teachers with diagnostics about students that will enable them to intervene as soon as help is needed, rather than when a student starts to struggle.</p><h3>Artificial intelligence government applications include emergency response</h3><p>If there’s one thing that has been brought to light by the 2020/2021 health crisis, it’s a worldwide lack of disaster preparedness. The need for a timely, coordinated response isn’t just imperative when it comes to pandemics. It also applies to large-scale weather events, natural catastrophes, geopolitical tensions and terrorist attacks alongside personal accidents and emergencies, whether at a local, national or global scale.</p><p>When disaster strikes, the ability to respond immediately, precisely and effectively is vital. At this point in time, the emergency notification process tends to be extensive and complex. When a person calls the emergency services to report an incident, a lot of valuable risk assessment time is lost on trying to understand exactly what the situation is. Callers in panic are not always able to provide objective, relevant information, while call quality and background noise can affect dispatchers’ ability to accurately evaluate a situation.</p><p>According to <a href="https://assets.documentcloud.org/documents/6985891/AI-Disaster-Resilience.pdf">recent Microsoft research into the California wildfires</a>, AI is one of the most effective ways to minimize the fallout from major incidents. The study states that “More adoption of AI could enable emergency management agencies to rally more precise responses to events more quickly, potentially saving lives and reducing property damage, or to cut costs on preventative capital improvement projects.</p><p>AI can help dispatchers identify the nature of an incident, how serious it is and flag a need to request additional information. This enables emergency center staff to ensure that the right kind of help is sent as soon as possible. The Estonian Emergency Response Center, which handles around a million calls per year, is <a href="https://mindtitan.com/resources/blog/mindtitan-is-developing-artificial-intelligence-assistant-for-estonian-emergency-response-center/">developing a solution</a> that uses AI to make risk assessment decisions faster and streamline emergency contact center operations.</p><p>Analyzing calls using text-to-speech and sentiment analysis enables the Estonian government to respond better to emergencies. But at the same time, it helps public agencies understand query patterns and take pre-emptive action to explain and handle issues.</p><h3>Using AI-based road traffic accident risk prediction to reduce accidents</h3><p>Similarly, a road accident prediction project is providing valuable information on how to reduce and prevent traffic incidents. The figures are sobering: in 2018, 67 people died and 1,832 were injured on the Estonian roads. The number of serious road accidents with victims has fluctuated between 1,300 and 1,500 since 2010.</p><p>When we think of AI and road traffic, it’s often in relation to self-driving vehicles. But AI has several applications when it comes to improving road safety. Towns across the world are implementing AI solutions to <a href="https://news.usc.edu/148660/usc-engineers-use-artificial-intelligence-to-reduce-traffic-jams/">monitor and optimize traffic</a> as part of “smart city” initiatives. AI can also help predict, prevent and reduce road accidents by analyzing the root causes of traffic accidents and what makes them more or less severe. Thus, cities can develop transport systems that enable sustainable mobility and shape safe attitudes and behaviour on the part of drivers. Over in Estonia, <a href="https://mindtitan.com/resources/blog/mindtitan-develops-an-artificial-intelligence-road-traffic-accident-risk-prediction-model-for-the-estonian-road-administration/">the Estonian Road Administration</a> is developing a forecasting model that predicts the risk, severity, and causes of traffic accidents. The solution will draw on multiple sources, such as traffic offenses, traffic accidents, weather conditions, locations of police vehicles, traffic count data and road register data.</p><h3>Conclusion</h3><p>According to the Estonian Government Chief Data Officer, Ott Velsberg, “<a href="https://www.youtube.com/watch?v=LiNLvuCAke4&amp;ab_channel=ITU">We need to rethink how to make the government as lean, as proactive and as personalized as possible</a>“. Estonia’s heavy AI investments are putting the country on the right path: AI for government can help the public sector design better policies, improve decision-making, and boost communication and engagement with residents. What’s more, it can improve the efficiency and all-round user-friendliness of services.</p><p>But AI can go beyond improving services and reducing bureaucracy. It also helps governments respond better to emergency situations, prevent them from happening in the first place and come to the aid of their citizens when they need it the most.</p><p>MindTitan works with public sector agencies to create innovative solutions for complex problems, providing both purpose-built solutions and tools from a developed product suite. Book a consultation now to find out how MindTitan can develop tailored solutions that fit the needs of both civil servants and citizens.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=a413860d0666" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[8 most important benefits of chatbots for business you need to know]]></title>
            <link>https://mindtitan-ai.medium.com/8-most-important-benefits-of-chatbots-for-business-you-need-to-know-ead10b1e3bad?source=rss-69684ed99bb2------2</link>
            <guid isPermaLink="false">https://medium.com/p/ead10b1e3bad</guid>
            <category><![CDATA[chatbots]]></category>
            <category><![CDATA[chatbot-benefits]]></category>
            <category><![CDATA[customer-service]]></category>
            <category><![CDATA[ai-chatbot]]></category>
            <category><![CDATA[ai]]></category>
            <dc:creator><![CDATA[MindTitan]]></dc:creator>
            <pubDate>Wed, 09 Jun 2021 07:23:04 GMT</pubDate>
            <atom:updated>2021-08-19T19:29:45.853Z</atom:updated>
            <content:encoded><![CDATA[<p>I doubt you have not heard about some cases where chatbots failed miserably in having any actual business benefit. Our experience working with the biggest companies in the world like Elisa, Veon, and Swedbank has taught us some interesting things we are about to share in this article.</p><p>The benefits of a chatbot can vary depending on the industry, type of chatbot, and size of the organization but commonly it saves time, money, and manpower. This article goes through the 8 most important benefits of chatbots that are essential for any organization that considers implementing one into the business.</p><figure><img alt="chatbot benefits" src="https://cdn-images-1.medium.com/max/954/1*3eY5vLJCP9n-2h37aNInlw.jpeg" /></figure><p>A chatbot is something that most people have interacted within their life or have heard at least on the concept level.</p><p>In fact, it has become an essential part of many businesses that have customer service, especially if it is a mature industry with high customer expectations and big competition on the market.</p><h3>1. Is available 24/7</h3><p>You probably have no idea how many users are browsing your website off-hours with questions about the service or a product. It might be your existing customer that would want to cancel their subscription or a potential customer that makes online research on a service. Some companies hire additional people to serve customers at later hours. However, it still does not mean that customers are ready to wait for their requests to be handled. In some industries, even one minute can mean that you either turn the potential client into a paid customer, or you never see them again.</p><p>The intelligent <a href="https://mindtitan.com/resources/guides/chatbot/">chatbot</a> does not have such issues as it can answer a big portion of customer inquiries. <a href="https://mindtitan.com/resources/blog/8-most-important-natural-language-processing-nlp-applications/">Natural language processing models</a> allow chatbots to resolve customers’ inquiries with a high first contact resolution rate. These models truly perform the task for a client without any human assistance. Even if you have a call center and customers are turning to you during off-hours, AI chatbots are able to handle these requests by implementing additional components from the <a href="https://mindtitan.com/products/call-automation/">call automation bot</a>. The speech recognition model transcribes the call in real-time and the NLP classifier identifies the issue behind the request. If a chatbot has an answer and the customer agrees to continue the conversation in the chatbot, an SMS with a chatbot link is sent to the customer, allowing to resolve the issue without human assistance.</p><h3>2. Creates personalized experience</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*TOEA7vRBPFJf32R0QoByDQ.jpeg" /></figure><p>Another important chatbot benefit that sometimes is overlooked relates to customer personalization. As revealed in the Segment research, 71% of the consumers are not happy when their shopping experience is impersonal. A chatbot is able to process customers’ personal data while browsing, which allows the bot to make a personalized suggestion or give advice during the session should the customer experience a problem.</p><p>Intelligent chatbots have the possibility to integrate with the backend systems through an API connection. This makes customers’ experiences more personalized as the bot can instantly analyze relevant background information. It will solve issues based on personal data. For example, instead of providing generic information about invoices, the bot looks at the person’s invoice data and communicates that. It can also make transactions instead of giving generic instructions on how to do these. Corporate chatbot use cases give a good overview on the capabilities of such enhanced systems.</p><h3>3. Detects sales opportunities</h3><figure><img alt="sales opportunities" src="https://cdn-images-1.medium.com/max/1024/1*PpcAh6X-829FuUM_wfBmFg.jpeg" /></figure><p>There are various ways how a chatbot can leverage sales revenue. One of these is the outbound chatbot that you probably have faced yourself on some websites. Such bots push messages with different options, e.g. time booking for a barber visit or purchase of a service that does not require human agent attention. These bots send messages while you are browsing a website or online store, and allow integrating sales funnels with lead magnets and follow-up messages. Such chatbots are limited to a specific function and may sometimes annoy customers.</p><p>Some enhanced AI chatbots are able to detect sales opportunities and forward these to the appropriate agent or department. It is crucial for big enterprises with thousands of inquiries a day. With long waiting time, some of the sales opportunities are lost, as customers are not willing to wait. For instance, Titan AI Chatbot software implemented at one of our customers from the telecommunication industry detects additionally around 5% of contacts with sales potential, 50% of which turn into paid customers. The AI chatbot can also automate the sales process if there is integration with the backend systems through an API and business decisions allow it.</p><h3>4. Cuts down costs</h3><p>Based on our experience and feedback from customers, on average cost per human customer service contact is EUR 3 to 5, depending on the market, the complexity of the questions, and the industry. The calculation can be performed by dividing the operational costs (OpEx) related to human agents by the number of contacts handled.</p><p>A truly intelligent chatbot can automate around 60% of all the customer contacts, out of which it can fully resolve around half. Depending on the case, the resolution rate can be increased even up to 70%. This means that with 5000 incoming customer inquiries per month and 60% bot automation, a company will save EUR 10,500 on costs. Of course these numbers may vary depending on the industry, specifics of the implemented chatbot and business operations of the company.</p><h3>5. Improves onboarding of new customer service agents</h3><p>Even the best AI chatbots automate only 60% to 80% of the incoming inquiries. Still some of the requests require human touch, for instance questions regarding termination of the contract or complicated technical issues. In big organizations like telecommunication companies or banks there might be hundreds of different questions that crop up. Knowing everything can be tricky even for experienced employees. Especially considering the fact that answers are written in manuals that are updated from time to time. Also, businesses should keep in mind the fact that customer support agents do not work longer than <a href="https://youtu.be/ZfUQHLElx2M?t=470">2 years in the same position</a>, meaning that companies have to constantly onboard and train new employees.</p><p>Some enhanced chatbots have a possibility to forward specific clusters of questions to certain employees. This does not only increase the speed of staff onboarding, but also the quality of the answer, as it is easier to become an expert in one question than dozens at the same time.</p><h3>6. Turns your CapEx into OpEx</h3><p>Nowadays, many companies switch from traditional hardware and software ownership to using SaaS business models. This helps to optimize capital expenditures (CapEx) and increase the amount of free cash flow that can be reinvested into other parts of the business.</p><p>When it comes to customer service and an increasing amount of incoming customer contacts, the strategy of building additional customer contact centers and hiring new agents is not efficient. It will require a heavy investment into the building and the infrastructure. Besides, if you rely on outsourcing your customer service, it is more difficult to control the quality of the service. As the Covid crisis showed, some companies have had to completely restructure customer service within one day.</p><p>In contrast, chatbot software technologies allow scaling the business fast in case you have a sudden increase in contacts. You can turn your CapEx into OpEx, as you will pay only for the contacts handled.</p><h3>7. Standardizes the solutions and creates a knowledge base</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*bxX3nb6U4pKuiWjuRZI01g.jpeg" /></figure><p>Human beings are predisposed to interpret an answer based on their personal understanding. In customer service, there are always people that give answers based on their feeling on what is right, as people tend to rely on their experience and are often overconfident. Even if most agents will follow manuals, it will not guarantee that the customer will get the right answer, as sometimes manuals are outdated. Without a proper solution, customer agents will either give a wrong answer based on their personal interpretation of the manual or simply due to the fact that it has not been updated on time.</p><p>A chatbot allows standardizing the knowledge base across all customer service. This almost fully eliminates the human error factor and guarantees meeting the expected SLA. For instance, <a href="https://mindtitan.com/products/chatbot-software/">Titan AI Chatbot software</a> uses a unique process of solution flow creation, which allows to instantly make any changes in the dialogues in case of some product updates. We define this unique process as product management 3.0 that allows the product owner to be in charge of the dialogue creation, which helps them to understand what are the actual problems that concern the client and how to solve them efficiently. Read our story about <a href="https://mindtitan.com/resources/blog/automated-customer-service/#how-to-actually-create-truly-automated-customer-service">truly automated customer service</a> to understand how chatbot can help with that.</p><h3>8. Give customer service cases an owner who is responsible for results</h3><p>When communication goes through many people (agent, customer service manager, and finally the person who deals with that part of the business), responsibility is shared. In this case, it becomes very complicated to understand what customers are actually asking and make improvements in the customer service procedures.</p><p>With chats, it’s easier for the final responsible person to review how the issue is handled and suggest or make improvements. This helps to standardize the process and ensure that provided information is up to date. For instance, Titan chatbot software uses solution flow management that allows product owners to be in charge of creating dialogues, to understand the customer’s reason to turn to customer support and to eliminate unnecessary contacts in the future. This precludes confusion and shared responsibility, and helps to focus on key goals in customer service.</p><h3>In summary</h3><p>The most important benefits of chatbots have been described in this article. Depending on the industry and specific use case, the biggest benefits from the business perspective come from the significant decrease in customer service cost, ability to detect sales opportunities, and optimization of the employee onboarding process. From the client’s perspective, a chatbot improves the customer experience and matches their expectations better because the answers are instant, just like with Google. If companies will not start investing in automation, they will be left behind.</p><p>We recommend reading our article about truly automated customer service that will give you a deeper understanding of the way you can automate customer service with enhanced AI systems. If you are planning to implement a chatbot in the near future, please keep in mind that you cannot treat it as a regular IT project. We have seen cases where companies fail due to wrong understanding of the process.</p><p><em>Originally published at </em><a href="https://mindtitan.com/resources/guides/chatbot/benefits-of-chatbots/"><em>https://mindtitan.com</em></a><em> on June 9, 2021.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ead10b1e3bad" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[AI Bot Annika Can Now Detect Network Anomalies and Service Errors]]></title>
            <link>https://mindtitan-ai.medium.com/ai-bot-annika-can-now-detect-network-anomalies-and-service-errors-fc482e1877ef?source=rss-69684ed99bb2------2</link>
            <guid isPermaLink="false">https://medium.com/p/fc482e1877ef</guid>
            <category><![CDATA[chatbots-for-business]]></category>
            <category><![CDATA[customer-service]]></category>
            <category><![CDATA[chatbots]]></category>
            <category><![CDATA[anomaly-detection]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <dc:creator><![CDATA[MindTitan]]></dc:creator>
            <pubDate>Thu, 29 Oct 2020 13:22:08 GMT</pubDate>
            <atom:updated>2020-12-01T13:01:52.599Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*U98JE75FbfbkXPfTaPiuvg.jpeg" /></figure><p>Up until now, Annika, the Elisa customer service bot, has been mainly used in the task of serving clients. Now, however, <strong>it has acquired new skills that help monitor Elisa’s network anomalies and service errors.</strong> The artificial intelligence listens to incoming calls, identifies customers’ problems, classifies the anomalies and informs specialists about likely failures and the probable locations of these on the map of Estonia.</p><p>AI has been used in Elisa’s customer service in the form of service bot Annika since 2018. “Annika started as an <a href="https://www.mindtitan.com/ai-customer-experience-telecommunications/">AI chatbot</a>. A year later, she also became a <a href="https://www.mindtitan.com/automate-your-whole-customer-service-not-only-chats/">callbot</a> to work in the call service, directing the client to the service staff member best equipped to solve the client’s issue. This year a third function was added. Now, Annika can provide real-time information about potential failures to Elisa’s monitoring service,” says Mailiis Ploomann, Head of Telecom Services at Elisa Eesti AS.</p><p>“The special situation that we were all faced within the second quarter of 2020 proved very well how good Elisa’s previously taken decisions had been. This encouraged us to continue to be bold in expanding Annika’s skills and features. Elisa offers a vital service to its clients and, therefore, needs to monitor the functioning of all services in real-time and at all times. By now we have learned to monitor hundreds of different IT applications with the help of special systems that allow us to detect failures and anomalies as quickly as possible,” explains Ploomann.</p><p>However, there are still situations that the used algorithms are unable to spot or that the previously agreed metrics cannot adequately reflect. In such a situation monitoring is best aided by AI that is quickly able to process hundreds or thousands of complaints in order to calculate what clients are unhappy about and where the problems lie.</p><p>Generally, the clients are known to inform Elisa if there is something wrong with a used service. “<strong>Because Annika answers all customer calls now, we have established a connection between AI and monitoring. At that, Annika uses various classifications to identify anomalies and to signal if a particular service is causing more issues than before</strong>,” Antti Suursalu, Leading Monitoring Specialist at Elisa, explains.</p><h3>Elisa’s AI and service bot transmits monitoring information in three categories:</h3><ol><li>Location — <strong>there’s a map view of incoming calls where each unique caller is depicted as a dot</strong>. The more dots in a region, the more precise the fault communication. Also, we are able to avoid treating the customer’s complaint as an individual case.</li><li>Problem — <strong>Annika transcribes the call and displays the information at the dot of each unique caller together with all other information Elisa knows about that client.</strong> Through that, Elisa is able to relay the contract or the call number to the specialist dealing with solving the problem. Additionally, the transcription enables to assess problem and its impact.</li><li>Service and network anomaly detection — for most calls, Annika is able to understand which service-related fault the customer wants to notify Elisa about. This enables to measure the number of incoming calls that concern a particular service and compare it to previous metrics. <strong>If the number of complaints differs remarkably from what it was at a compared time period, the bot informs a specialist about the anomaly.</strong></li></ol><p>Customer service bot Annika is AI that has been developed in cooperation between Elisa and <a href="https://www.mindtitan.com/">MindTitan</a>. It has been in use since 2018, whereas Annika started to take client’s calls in May 2019. Markus Lippus, Leading Data Scientist at MindTitan, explains the difficulties of speech recognition in telephone calls, “Firstly, archived phone calls have very low sound quality, and secondly, these are full of various kinds of noise which can be caused by the specifics of the telephone or the way the signal is transmitted or anything that may cause background noise. These reasons make work harder for the acoustic model — a component that separates human vocals from the sound stream. The third problem is that people speak very differently from one another and also imperfectly — some only pronounce half of the word, use synonyms or general terms instead, or even skip words.</p><p>All of this complicates the work of the language model that has the task of putting the identified sounds into words and sentences. “People are able to understand such speech because we spend a large part of our lives distinguishing human speech from other sounds. We have extensive knowledge of the world and how it works. This helps us to assume from context which word is missing from the sentence, what was mispronounced or what was only hinted at.”</p><p>Adding complexity to failure detection are two more models: the classification model, which must understand very different problem descriptions to understand exactly where the error might be; and the anomaly detection model, which assesses the content, volume and area of calls to see if the problem is pointing to a particular device or a whole area. “These steps are crucial, because if calls are transferred incorrectly or a service is misdiagnosed, it is easy for regional failures to go undetected or for some services to be given false alarms which waste time,” Lippus adds.</p><p>Service bots are one of the most common <a href="https://www.mindtitan.com/types-of-chatbots/">types of practical application of AI</a> in the world. Only a few years ago, Annika was one of the few of its kind customers came into contact with, but <strong>AI chatbots</strong> are becoming more and more common in most major service environments. “Callbots are the next wave of developments in Estonia and Annika is again a pioneer in the field,” Ploomann notes.</p><p><a href="https://www.elisa.ee/en/ai">Elisa Eesti AS</a> belongs to the Elisa Group, one of the leading Telcos in Nordics.</p><p><em>Originally published at </em><a href="https://www.mindtitan.com/network-anomalies-and-service-errors/"><em>https://www.mindtitan.com</em></a><em> on October 29, 2020.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=fc482e1877ef" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[8 Most Important Natural Language Processing (NLP) Applications]]></title>
            <link>https://mindtitan-ai.medium.com/8-most-important-natural-language-processing-nlp-applications-fe47a10a8aae?source=rss-69684ed99bb2------2</link>
            <guid isPermaLink="false">https://medium.com/p/fe47a10a8aae</guid>
            <category><![CDATA[application]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[for-businesses]]></category>
            <category><![CDATA[business-operations]]></category>
            <category><![CDATA[nlp]]></category>
            <dc:creator><![CDATA[MindTitan]]></dc:creator>
            <pubDate>Mon, 19 Oct 2020 13:56:35 GMT</pubDate>
            <atom:updated>2020-12-01T13:06:53.143Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/640/1*PDQbZYAqMg0SyAe06DJrRA.png" /></figure><p>The amount of information and the number of interactions online is growing exponentially, ranging from social media content to the never-ending stream of customer interactions. The need to access and synthesize text in real-time or as soon as it is posted becomes increasingly important. Businesses can’t keep up with this never-ending stream of information by simply hiring more customer service representatives. Instead, they must incorporate <a href="https://en.wikipedia.org/wiki/Natural_language_processing">natural language processing </a>(NLP) technology into their operational workflow.</p><p>NLP is a field operating at the intersection of artificial intelligence, computer science, and linguistics. NLP applications are capable of solving a wide set of text, syntax, semantics, and speech problems. NLP applications decipher text by simulating the human ability to understand the language.</p><h3>What will you learn?</h3><p>In this article you will:</p><ul><li>learn about 8 most important NLP applications</li><li>understand how you can use them in your everyday business operations</li></ul><p>Let’s dive in together to understand the 8 most relevant NLP applications and explore how can they help you become more profitable.</p><h3>1) Text classification</h3><p><a href="https://developers.google.com/machine-learning/guides/text-classification">Text classification </a>is the core task in natural language processing.<strong> The goal of text classification is to read the text and assign one label from a fixed set of categories to it based on its content and detected keywords.</strong> Despite its simplicity, text classification is often the first step in many natural language processing pipelines. The reason is that many businesses can’t extract valuable insights from unstructured text data, such as emails, news, customer interactions, or investor outlook, before the data is categorized in a certain way. And this is where text classifiers shine.</p><p>The text classifier is a supervised learning algorithm that quickly and accurately sorts the text into categories or topics on which the algorithm was trained on. Let’s illustrate this with an example. Imagine that you want to sort all news headlines or Twitter feeds into 3 categories: 1) technology, 2) sports, and 3) fashion. In the training phase, you feed these news together with their correct labels into the algorithm. In the prediction phase, you apply this algorithm on the new unseen Twitter feeds and the algorithm will predict its topic.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/700/0*PLSq2yS9Cz5VaunS.png" /><figcaption><a href="https://towardsdatascience.com/automated-text-classification-using-machine-learning-3df4f4f9570b">Source</a></figcaption></figure><p>Text classifiers can be applied on a variety of textual datasets, be they emails, tweets, news headlines, or news articles. <strong>Overall, text classifiers can predict topic, sentiment, intent, emotion, or perform any kind of classification you want the algorithm to be trained on</strong>. Below, we also explore <a href="https://en.wikipedia.org/wiki/Sentiment_analysis">sentiment analysis </a>in more detail.</p><h3>2) Sentiment analysis</h3><p>Sentiment analysis is a subfield of text classification. <strong>Its goal is to process the text to classify the author’s sentiment towards something.</strong> This sentiment can be classified into categories, such as positive, neutral, or negative, or into broader response groups, such as anger, disappointment, fear, or satisfaction. Popular use cases of sentiment analysis include advertising, brand management, finance, or trading.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/850/0*0nUobJYUNoub1gb7.png" /><figcaption><a href="https://www.researchgate.net/figure/Example-of-sentiment-analysis-at-different-levels_fig2_283664122">Source</a></figcaption></figure><p>For instance, many banks, cryptocurrency exchange platforms, and portfolio management companies use market sentiment analysis to predict the trend in stock or cryptocurrency prices. Market sentiment is an aggregate measure of market opinions, moods, and outlooks that investors voice through social media or microblogging websites. Portfolio management companies agree that market sentiment is highest just before the market reaches high and lowest just before the market reaches <a href="https://catanacapital.com/blog/sentiment-analysis-stock-market-sentiment/">rock-bottom </a>. As a result, market sentiment analysis creates attractive investment opportunities for investors before there is any news to explain this market movement.</p><h3>3) Machine translation</h3><p>Machine translation systems are extremely helpful for any business with global operations or global customer reach. <a href="https://www.babbel.com/en/magazine/how-many-people-speak-english-and-where-is-it-spoken">Linguistic studies </a>show that out of the world’s 7.5 billion inhabitants, only 5% speak English as their first language and only additional 15% have some degree of English comprehension. Therefore, if you are relying only on English or your company’s native language, you are missing out on a significant fraction of your potential customer or client base. Machine translation breaks down this barrier by quickly and accurately translating catalogs, reports, technical manuals, or customer interactions.</p><p>The cutting-edge machine translation algorithms also include localization which addresses a variety of semantics and contextual factors, such as local idioms, cultural references, measurement units, date formats, and page sizes. An example of localization is translating from US English to French where imperial measurements (feet, inches, etc) are converted to metric units (meter, centimeters). Similarly, dates will be localized as well. For instance, 1/12/2020 indicates January 12th in the US but will be translated to 12.1.2020 in French to preserve the correct meaning.<strong> The cutting-edge technology behind machine translation systems are deep learning networks, which translate the words in the source text based on the localization and context of that word used in the sentence.</strong></p><h3>4) Bias and fake news detection</h3><p>Digital misinformation and fake news have been declared as the major technological and socio-political risk by the <a href="https://scholar.google.com/scholar_lookup?&amp;title=Digital%20wildfires%20in%20a%20hyperconnected%20world&amp;journal=WEF%20Rep.&amp;volume=3&amp;pages=15-94&amp;publication_year=2013&amp;author=Howell%2CL">World Economic Forum </a>. <strong>Fake news are created with malicious intent to misinform readers, generate unnecessary polarizations among opposing groups, and drive traffic to news with clickbait headlines or politically biased content.</strong> Recent examples include climate change propaganda or 2016 presidential elections in the US where around 100 million of Facebook accounts were targeted by fake social messages backed by Cambridge Analytica. As an illustration, Facebook AI team recently released a natural language processing model called , which was later transformed to detect whether the news is real or fake based on the localization and semantic context of tokens it contains.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*yrmg9lkBFLxxxoor.png" /><figcaption><a href="https://www.analyticsvidhya.com/blog/2019/12/detect-fight-neural-fake-news-nlp/">Source</a></figcaption></figure><h3>5) Named entity recognition</h3><p><a href="https://www.kdnuggets.com/2018/08/named-entity-recognition-practitioners-guide-nlp-4.html">Named entity recognition</a> is a natural language processing application which identifies named entities from a text and assigns them into pre-defined categories. Named entities can be names of people, universities they attended, companies they worked at, dates, and quantities.<strong> Entity identification is extremely valuable for businesses that require fast and scalable analysis of long and unstructured text documents.</strong> One prevalent real-world use case is an <a href="https://ner.snapxplatform.com/">analysis of applicants’ resumes </a>by HR departments. Your HR department probably receives thousands of applications for a single position and spends countless hours wading through candidates’ profiles to find the perfect match. Each resume is formatted differently, yet contains the same type of information, such as the candidate’s name, universities they graduated from, or previous companies. Deploying entity identification enables your recruitment teams to immediately extract critical information, classify it into your pre-defined categories, and create a uniform snapshot that lets you evaluate multiple candidates instantly.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/852/0*QYM2nlXoh_MduReq.png" /><figcaption><a href="https://ner.snapxplatform.com/">Source</a></figcaption></figure><h3>6) Predictive text modeling</h3><p>Predictive text modeling or next word prediction is an extremely useful NLP application deployed in messaging applications, such as WhatsApp, email, Google Search, and chatbots. The algorithm learns to predict the next word in a sentence by understanding the localized context from that sentence. As an illustration, when I type <em>how to learn </em>in the Google Search box, I already see some predictions. I see that certain next words are predicted given the existing words <em>how to learn. </em>The autocomplete or next word prediction saves a lot of time for your users and helps them express their needs by understanding their patterns of writing and background.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*Z8WY2_hv61SH6iUH.png" /><figcaption>Google</figcaption></figure><h3>7) Chatbots</h3><p>A chatbot is a scalable and efficient solution to automate your customer service, recruiting, and marketing workflow. <strong>The chatbot is a starting point for a conversation with your website visitors. </strong>It finds out what the visitors are searching for before they drop off. One of the key advantages of chatbots is that they replace long waiting times when a customer has an inquiry with real-time, precise answers 24/7. This can significantly boost your conversion rates, help you nurture clients, and generate revenue. It also prevents important clients dropping off or slipping through due to missing their message or not responding quickly enough.</p><p>Chatbots are relatively easy to implement with very little input from you, such as pre-defined responses. In turn, they significantly cut down your costs by automating your workflow and escalating to the right people on your team with more advanced queries. It is a win-win scenario. But there are different <a href="https://www.mindtitan.com/types-of-chatbots/">types of chatbots </a>you can implement into your business depending on your use case, one of which is the AI chatbot.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/331/0*xVgdLLGG0a7vR0x7.gif" /><figcaption><a href="http://www.mindtitan.com/">AI Chatbot</a></figcaption></figure><h3>8) Recommender systems</h3><p>Content-based recommender systems are another popular NLP application that provides delightful user experiences and creates sales funnels. Many large retailers, such as Zalando, Amazon, eBay, and Netlix, leverage recommender systems. Content-based recommender systems rely on two approaches: item-item similarity and user-user similarity. We are going to focus on item-item similarity because it requires less specific knowledge about a particular visitor.</p><p>As an illustration, all clothing products sold by fashion giant Zalando are labelled with hundreds of attributes, such as color, body shape suitability, price, length, occasion, season, and gender. <strong>Whenever a visitor buys something from Zalando, let’s say grey jeans, other articles that match those jeans’ attributes will be automatically recommended.</strong></p><p>The NLP algorithm relies on similarity metrics, such as Euclidean distance or correlation similarity, to tell us how closely two articles with all their attributes are correlated. Zalando reports that outfit recommendations drive<a href="https://www.forbes.com/sites/bernardmarr/2019/09/20/the-amazing-ways-retail-giant-zalando-is-using-artificial-intelligence/"> 40% larger purchases. </a>All in all, recommender systems help customers effortlessly explore similar outfits while driving incremental revenue.</p><p><strong>Next steps</strong></p><p>We’ve gone through the 8 most important NLP applications and outlined the ways they can help you automate your workflow. But please don’t stop here. Our team has extensive experience with designing all of these systems and is available to work with you on the best solution for your specific needs. Reach out today to get started on this exciting journey!</p><p><em>Originally published at </em><a href="https://www.mindtitan.com/8-most-important-natural-language-processing-nlp-applications/"><em>https://www.mindtitan.com</em></a><em> on October 19, 2020.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=fe47a10a8aae" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Types of Chatbots. Rule-Based Chatbots vs AI Chatbots]]></title>
            <link>https://mindtitan-ai.medium.com/types-of-chatbots-rule-based-chatbots-vs-ai-chatbots-f4d2b7cc818b?source=rss-69684ed99bb2------2</link>
            <guid isPermaLink="false">https://medium.com/p/f4d2b7cc818b</guid>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[chatbots]]></category>
            <category><![CDATA[business-value]]></category>
            <dc:creator><![CDATA[MindTitan]]></dc:creator>
            <pubDate>Tue, 29 Sep 2020 08:24:33 GMT</pubDate>
            <atom:updated>2020-12-01T13:06:30.629Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/640/1*9OAHvidAXSFU7Fc4BNpB3g.jpeg" /></figure><p>The global chatbot market is going to reach <a href="https://markets.businessinsider.com/news/stocks/global-chatbot-market-anticipated-to-reach-9-4-billion-by-2024-robust-opportunities-to-arise-in-retail-ecommerce-1028759508">$9.4 billion by 2024</a>. Hundreds of companies worldwide develop various types of chatbots that mostly intend to help businesses to improve their clients’ experience and decrease customer service costs. A “big market pie” motivates companies to develop new solutions. However, some of these are fairly outdated to serve actual business needs in 2020. This article will help you understand what the different types of chatbots are, what these are used for, and which ones could have the biggest value to you.</p><h3>Rule-based chatbots vs AI Chatbots</h3><p>Rule-based chatbots became very popular after <a href="https://techcrunch.com/2016/04/12/agents-on-messenger/">Facebook launched</a> its Messenger platform where chatbots allowed businesses to perform automated customer support. Such types of chatbots are used for answering simple questions, for instance, when it comes to booking a table in a restaurant, buying tickets to the cinema, or using online delivery services. Guided by a decision tree, the customer is given a set of predefined options that lead to the desired answer.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*Q8JRP_lnvAaLUkVJ.jpg" /></figure><p>Rule-based chatbots are often split into two tracks: a sales track for capturing contact details, setting up a call or a meeting; and a support track for giving generic answers or sending a website link containing the necessary information. In most cases, these types of chatbots are built with a graphical interface reacting to the user pressing a specific button that activates the next layer of the decision tree. Often, these types of chatbots are based on keyword rules, reacting on specific words but these are limited to typos and have a risk of providing wrong answers or causing very frustrating customer experiences.</p><p>AI chatbots, on the other hand, use natural language processing (NLP) technologies to understand the intent behind the question and solve the customer’s problem without any human assistance. The biggest difference with the rule-based chatbot is the usage of the machine learning models that significantly increases the functionality of the bot as it is able to identify hundreds of different questions written by a human.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*1o0fPDmF3fSODI6o.jpg" /></figure><p>A customer can ask the bot a question by writing it in the same way as if asking a human agent. The bot uses a text classifier to identify the intent and understand the meaning behind the question. The next step involves creating a sequence of additional questions and answers using the dialogue tree that helps to specify the exact issue the customer wants solved and the way it will be done. Besides that, API integration with the back end systems allows the bot to really perform the task for the customer, instead of just providing a link to self-service instructions. NLP models help an AI chatbot to identify hundreds of different questions making it very useful not only for big enterprises in branches like <a href="https://www.mindtitan.com/case/ai-in-telecom-business/">Telecommunication</a>, <a href="https://www.mindtitan.com/case/artificial-intelligence-in-finance-and-banking/">Finance and Banking</a> sectors, but also the Public sector, as it is able to serve thousands of customers automatically and perform complicated tasks on its own.</p><h3>Are NLP-based types of AI Chatbots better than rule-based ones?</h3><p>It is not right to say that rule-based chatbots are worse, they are just different. They are much simpler, have a specific usage and are more affordable. In contrast, AI chatbots are used for more complicated cases to fully resolve customers’ issues. Also, rule-based bots are limited by typos or wrong keywords that people might use. Another problem with rule-based chatbots is that there is a limited amount of questions that such a bot can handle because it uses decision trees; it has a limited amount of questions that a client can navigate through. Clients can get lost when it comes to more complicated services than just ordering a pizza. And finally, NLP technology has become easily accessible for every company independent of its size. This is why further in this article, we are going to discuss the business value of various AI chatbot implications that are based on NLP technology.</p><h3>The business value of different types of AI chatbots</h3><p>AI chatbots have a more significant value, as they can automate a bigger number of questions, require less data for training (per each question it is trained to understand) and can solve more complicated issues compared to the rule-based chatbots. However, there are also different implications of the AI bot depending on the complexity of the use case.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*TblVYDgT456hFweL.jpg" /></figure><p>The given illustration is based on the actual experience of one of our biggest Telco clients in the Nordics — Elisa, that shared their AI <a href="https://www.mindtitan.com/ai-customer-experience-telecommunications/">chatbot use case</a> and compared <a href="https://www.callcentrehelper.com/first-contact-resolution-definition-formula-best-practices-149755.htm">first contact resolution rate (FCR)</a> in different project stages. As for their experience, a basic bot that uses NLP and handles simple requests, provides only 4% of the total business value that an AI chatbot can provide in maximum. Basic type of AI bot understands the meaning behind the questions, answers simple ones and sends informational messages, for example, a link to the FAQ section on the website.</p><p>The biggest business value comes from a personalized and transactional type of chatbot that can personalize messages and perform the tasks automatically for the customer. Customer verification and API integration with knowledge bases allow the bot to see client’s historical data and better understand their specific problem. Thus the bot’s response is personalized and it can even perform a task for the client, e.g., activate roaming. Also, using improved NLP algorithms allows the bot to reach 92% text classification accuracy (e.g., the bot classifies correctly 9 out of 10 client questions that the bot has been trained to understand) by confirming the intent with the customer. If a customer asks a question and the bot does not have a high enough confidence level to make a prediction on the client’s intent, the bot asks a specifying question, e.g., “Would you like to change billing information or personal information?”</p><h3>Additional Models</h3><p>There are several additional AI models that can be integrated into a chatbot so that it makes the bot much more powerful and intelligent. Experimenting with other models like next-best-offer (NBO) adds a possibility to recommend the customer the best matching products or services based on either previous interactions with products or services, user profiling based on similar customer segments, or behaviour on the website. From experience, this way an AI chatbot gives an additional 10% of business value.</p><p>The last step is the implementation of natural language conversations. From our experience, this influences business value the least, as businesses are not ready to pay extra for a feature that customers simply do not need. Clients do not value a natural and friendly conversation with a bot, they simply want their problem to be solved as soon as possible.</p><h3>Conclusion</h3><p>According to <a href="https://www.businessinsider.com/chatbots-vs-humans-for-customer-relations-2016-12?IR=T">Businessinsider</a>, 44% of US customers prefer chatbots over humans in customer service. This trend is in line with the habit of instantly finding information by just Googling it. When it comes to customer service, people have the same expectations. As technology continues to improve more and more users will be willing to use chatbots in their everyday life as it helps to save time and solve their issue without delay.</p><p>This article aimed to help understand the 2 main types of chatbots: rule-based and AI chatbots. The latter have a much more complicated functionality that requires less training data and that can actually perform the task for the customer without any human assistance. Also, there are different ways for implementing the AI bot depending on the use case. Hopefully, now you know enough and are ready to implement one of the bots in your own business. <a href="https://www.mindtitan.com/contact/">Book a Demo</a> with us to learn more about our AI chatbot software.</p><p><em>Originally published at </em><a href="https://www.mindtitan.com/types-of-chatbots/"><em>https://www.mindtitan.com</em></a><em> on September 29, 2020.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=f4d2b7cc818b" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Client Story: How AI and Callbot Helped Elisa to Automate Customer Service During The Crisis —…]]></title>
            <link>https://mindtitan-ai.medium.com/client-story-how-ai-and-callbot-helped-elisa-to-automate-customer-service-during-the-crisis-7dec24c0818c?source=rss-69684ed99bb2------2</link>
            <guid isPermaLink="false">https://medium.com/p/7dec24c0818c</guid>
            <category><![CDATA[customer-service]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[call-center]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[telecommunication]]></category>
            <dc:creator><![CDATA[MindTitan]]></dc:creator>
            <pubDate>Mon, 01 Jun 2020 11:06:24 GMT</pubDate>
            <atom:updated>2020-06-01T11:06:24.994Z</atom:updated>
            <content:encoded><![CDATA[<h3>Client Story: How AI and Callbot Helped Elisa to Automate Customer Service During The Crisis — From a Slow Cruiser to an Agile Beast</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*XAKnyaGloQegiGqZhkZ5cw.jpeg" /><figcaption>Mailiis Ploomann, Head of Telecom Services at Elisa Eesti AS</figcaption></figure><p>Only a few months ago to automate customer service seemed for many companies just a nice little addition. Today, however, we know to have entered a completely new era of social organization. In this new era, we should be ready to expect the unexpected, fast changes and the need to reorganize ourselves. In that new normal the winners are those who adapt well.</p><p>Service robots communicating in text are no longer a rarity in Estonia. In fact, they are becoming more and more common, even if many of these just answer FAQs or give a content overview. This has less to do with the robots capabilities than the fact that the owner is still at the very beginning of the learning curve.</p><p>The special conditions applied due to the pandemic shook all entrepreneurs. Having as good as the whole society move into their homes changed the external environment for <a href="https://www.elisa.ee/">Elisa</a> considerably. First of all, this meant a change for network use. It was like New Year’s Eve, when everybody wants to make a call at the same time, only with the difference that one night in the year turned into a daily 12-hour workday. Also, the number of people who needed Elisa’s help increased by as much as 20%. In addition to all that, we as a company also needed to relocate our 1000 employees to work from their homes. Thinking about it now this sounds like the perfect recipe for total anarchy, overall confusion and chaos. Plague, famine and the solar eclipse all on one day.</p><p>As crazy as that was, in practice, the chaos settled fast — every particle knew exactly what their individual task was. The whole company changed its form as if it were a transformer. What used to cruise comfortably became an agile warrior beast. The main reason for achieving that was conscious people management and an earlier practice of adjusting to changes. However, when it came to serving our clients, our <a href="https://www.mindtitan.com/how-an-ai-powered-call-center-automation-is-helping-elisas-customer-support-during-the-current-crisis/">customer service robot Annika</a> had an important role to play in reorganizing the work of our call center.</p><p>The call center deals with the clients that call Elisa. In an ideal world, each caller would be immediately put through to that service staff member that can best address their problem. According to our most conservative estimate, there are over 800 topics that our clients could be contacting us about. This is why it is impossible that any one person could be able to help each caller. As in many other larger companies, also in Elisa service staff members are specialized on a particular topic and in case they are approached with a “wrong” issue, they will forward the call to an employee who can best help with that particular problem.</p><p>Having a call forwarded like that is not nice for anybody. The representatives are not happy that they waste time on something that is not directly helpful for the client. The clients are not happy because they feel their time is being wasted. This is why in 2019 Elisa introduced a customer service robot, Annika, who answered all calls, listened to the client, and then transferred the calls to the right employee. A year ago our clients were still sceptical about the Estonian-speaking robot, thinking that they have to be very careful about which words or code names they use when explaining what their problem is. Today, just before Annika’s first birthday, clients are all very calm and natural when talking to the CallBot. There seems to be no difference to how they would talk to a human representative.</p><p>Now in March 2020 such an effective robot call transferring helped us in many ways. We were able to save our employees’ time by reducing the number of calls they had to forward to a colleague, and each client was served faster. Additionally, we were also able to register the problems of those clients who didn’t even get through to a service staff member due to the extraordinarily high demand. If clients themselves choose the right topic for their call in about 75% of the cases, then our April check-up found that Annika’s rate is 90%. Serving tens of thousands of clients a day it makes a huge difference whether we have to forward 10 or 25% of our callers.</p><p>What’s more, Annika enables an easier introduction period for a new employee who might not yet be able to answer all questions fully. In a situation where all of us are working from home, the learning process is, of course, slower than usual. Therefore, any kind of automated pre-filtering is even more important.</p><p>We are to expect similar sudden situation changes also in the future. We have all learned by now — or are in the process of learning — how to work from home. But in order to be really ready for the next crisis, we must have other solutions to go with it. People need a little time to adjust to changes and reorganizing our personal lives also requires attention. Each minute that we can save for our employees or each activity that can be simplified using AI is worth it. The sooner we can provide that, the better.</p><p>Therefore, if your government institution or private organization hasn’t already started to build a crisis-proof solution, know that the necessary technology already exists. It only needs to be taken into use.</p><p>Written by Mailiis Ploomann, Head of Telecom Services at Elisa Eesti AS</p><p>Have a specific machine learning project in mind? <a href="https://www.mindtitan.com/contact/">Contact us</a> for a free consultation.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=7dec24c0818c" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[MindTitan Develops an Artificial Intelligence Model for The Estonian Tax and Customs Board to…]]></title>
            <link>https://medium.com/digital-diplomacy/mindtitan-develops-an-artificial-intelligence-model-for-the-estonian-tax-and-customs-board-to-24289df87ae2?source=rss-69684ed99bb2------2</link>
            <guid isPermaLink="false">https://medium.com/p/24289df87ae2</guid>
            <category><![CDATA[tax-fraud]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[government]]></category>
            <category><![CDATA[public-sector]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <dc:creator><![CDATA[MindTitan]]></dc:creator>
            <pubDate>Tue, 26 May 2020 06:38:56 GMT</pubDate>
            <atom:updated>2020-05-27T14:41:27.149Z</atom:updated>
            <content:encoded><![CDATA[<h3>Developing an Artificial Intelligence Model for The Estonian Tax and Customs Board to Reduce Tax fraud</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*tX0-o4larTTbxx-DHeV0lg.png" /></figure><p><strong>The project aims to find an artificial intelligence model that would make it easier to identify potential salary tax fraud. To make this possible, </strong><a href="https://www.mindtitan.com/"><strong>MindTitan</strong></a><strong> is working on creating a model that can analyze and recognize patterns — mainly through tax return information — and as a result, indicate which companies should be more closely scrutinized.</strong></p><p>According to calculations by the <a href="https://www.emta.ee/eng">Estonian Tax and Customs Board</a> (ETCB), in 2019 the state lost 134,1 million euros of potential tax revenue due to so-called “under the table wages.” Taking into account that Estonia has a population of 1.3 million, as well as its advanced digital systems, the number is still extreme — something needs to be done.</p><p>Currently, ETCB makes decisions for identifying the sectors and companies that require attention, based on data analysis. This analysis is based on a series of declarations that companies are required to present each month. When the submitted declarations do not meet the requirements, the ETCB carries out a risk analysis.</p><p>The current risk analysis is rule-based, which means that rules for storing, sorting, and processing data are hand-crafted. Companies that are in risk groups are prioritized according to how high or low the risk is. The initial selection is made automatically and then reviewed by analysts; the final choice is made by an auditor, who also decides how to proceed.</p><p>According to Pille Muni, Development Specialist at the Tax Audit Department at ETCB, the problem with the old approach is that finding instances of fraud with such a system is very time-consuming and resource-intensive. Furthermore, the current model is not effective enough to achieve the desired result. Thus, the primary goal of the project is to find a solution that would solve this problem and allow for more efficient tax collection.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*M7qY-aYuLoe_I6cfohmbTQ.png" /><figcaption>Data is collected from multiple sources, and keeping humans in the loop helps AI learn to make higher-quality decisions.</figcaption></figure><p>MindTitan’s vision is to create a solution in which accuracy can be increased over time by building a human-in-the-loop system.</p><p>In a system with human supervision, artificial intelligence is learning to make human-like decisions while also finding connections that are sometimes overlooked by humans and that are not realistic to cover with a rule-based system alone. In the scope of the current project, MindTitan will use the existing data and potentially create additional data during the project.</p><p>The follow-up project would be to build an entire information system and data conveyors in order to automatically collect more training data and to train the system more intelligently. The aggregation of additional data sources in the follow-up project will also remain in place, but the data processing system used in the scope of the project will create preconditions for merging more and more data in the future, and for processing these efficiently.</p><p>As a result, the selection of companies that warrant closer scrutiny should become more accurate, and the probability of wages being paid under the table should decrease. This will also make the work of both analysts and auditors more efficient. It is possible that artificial intelligence will become smart enough in the future that the intermediate selection of objects by the analyst will become unnecessary in some cases, but the final decision will always be made by human officials.</p><p>Have a specific machine learning project in mind? <a href="https://www.mindtitan.com/contact/">Contact us</a> for a free consultation.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=24289df87ae2" width="1" height="1" alt=""><hr><p><a href="https://medium.com/digital-diplomacy/mindtitan-develops-an-artificial-intelligence-model-for-the-estonian-tax-and-customs-board-to-24289df87ae2">MindTitan Develops an Artificial Intelligence Model for The Estonian Tax and Customs Board to…</a> was originally published in <a href="https://medium.com/digital-diplomacy">Digital Diplomacy</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Developing Artificial Intelligence Assistant for Estonian Emergency Response Center]]></title>
            <link>https://medium.com/digital-diplomacy/developing-artificial-intelligence-assistant-for-estonian-emergency-response-center-13398e1de684?source=rss-69684ed99bb2------2</link>
            <guid isPermaLink="false">https://medium.com/p/13398e1de684</guid>
            <category><![CDATA[government]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[contact-center]]></category>
            <category><![CDATA[estonia]]></category>
            <dc:creator><![CDATA[MindTitan]]></dc:creator>
            <pubDate>Tue, 19 May 2020 13:23:42 GMT</pubDate>
            <atom:updated>2020-05-27T14:40:49.738Z</atom:updated>
            <content:encoded><![CDATA[<p><strong>The project aims to determine whether and how artificial intelligence can help make risk assessment decisions faster and save valuable time for the emergency contact center.</strong></p><p><a href="https://www.112.ee/en">The Estonian Emergency Response Center</a> is a government agency that is the first line of help in an emergency that requires the police, ambulance or a rescue unit. In case of emergencies the Center directs the ambulance or rescue workers to those in need. The Center handles about a million calls per year.</p><p>The current problem the Center is facing is that the emergency notification process is very extensive and complex. This means that a lot of valuable time is taken up during the risk assessment by the dispatchers trying to understand the situation.</p><p>The challenge in providing a risk assessment lies in the fact that during an emergency call, the dispatcher must understand the situation as soon as possible while taking into account the information needs of at least three different authorities. Gathering such information in limited time and with limited resources is difficult and time consuming.</p><p>This is where artificial intelligence could help. According to the Center, the primary goal is to see if it is possible to create a prototype model using AI that could identify what kind of incident has happened, how serious this incident is, and help the dispatcher identify if it is necessary to ask for additional information to make sure that the most appropriate help is sent.</p><p>From the technical aspect, AI will listen to and transcribe a real-time emergency call while taking into account the specifics of a call, like an environment noise, caller anxiety or diction, intermittent call quality, etc. The AI solution listens to and analyses the calls and makes suggestions for dispatchers but does not interfere directly with them, so nothing will change for the caller. The final decision will still be made by the dispatcher in terms of the type and priority of the event.</p><p>The solution will improve over time through machine learning, and ultimately reduce errors, and enable to save valuable time in aiding to send help faster.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*p8vZMqq7WW9jKcNwP79RPw.jpeg" /></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=13398e1de684" width="1" height="1" alt=""><hr><p><a href="https://medium.com/digital-diplomacy/developing-artificial-intelligence-assistant-for-estonian-emergency-response-center-13398e1de684">Developing Artificial Intelligence Assistant for Estonian Emergency Response Center</a> was originally published in <a href="https://medium.com/digital-diplomacy">Digital Diplomacy</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Client Story: Artificial Intelligence Helps to Completely Restructure Business Processes in Just…]]></title>
            <link>https://mindtitan-ai.medium.com/client-story-artificial-intelligence-helps-to-completely-restructure-business-processes-in-just-8147cff5f0c9?source=rss-69684ed99bb2------2</link>
            <guid isPermaLink="false">https://medium.com/p/8147cff5f0c9</guid>
            <category><![CDATA[customer-service]]></category>
            <category><![CDATA[covid19]]></category>
            <category><![CDATA[telecommunication]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <dc:creator><![CDATA[MindTitan]]></dc:creator>
            <pubDate>Fri, 15 May 2020 11:32:25 GMT</pubDate>
            <atom:updated>2020-05-15T11:32:25.386Z</atom:updated>
            <content:encoded><![CDATA[<h3>Client Story: Artificial Intelligence Helps to Completely Restructure Business Processes in Just Over a Day</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1000/1*WHFKtgYav-M5TPfhLcRObg.jpeg" /><figcaption>Mailiis Ploomann. Photo: Jake Farra, Elisa</figcaption></figure><p><strong>The current pandemic hit all companies equally unexpectedly. Like all others, </strong><a href="https://www.elisa.ee/en"><strong>Elisa Eesti AS</strong></a><strong> had to react fast in order to enable nearly a thousand employees to work remotely. At that, the company had to continue all of its business processes without making any concessions to its customer support or service quality. It has taken the past ten years of continuous work to perfect these business processes. However, because of the crisis, we had to turn everything upside down in a mere day and a half. Without the help of artificial intelligence, that would have been impossible.</strong></p><p><a href="https://www.mindtitan.com/ai-chatbot-telecommunications/">Customer service robot</a> <a href="https://www.elisa.ee/en/ai">Annika</a> that has been developed in Estonia in cooperation with MindTitan has been working for Elisa already since 2017. Over time the robot has been getting more and more independent in servicing actual clients. When special conditions were applied due to the pandemic in the spring of 2020, we learned that Annika is not only a platform for service automation, but also a foundation for agile activities in providing our services.</p><p>From the client’s point of view, Elisa has operated as a fully traditional telecom company or any large service company for decades. Our service scopes are large — we make hundreds of thousands of contacts in each calendar month. A client may call Elisa or write to the company and we also call and write to our clients. We have a respectable e-channel and up until the middle of March it was possible to visit us at our service points in shopping centers. This set of service channels was a natural part of most our processes and it enabled to criss-cross between contact points. For example, a client would have been able to agree the purchase of a product or a service on the phone, sign for it digitally, and then pick up the goods (e.g. their new phone) from a service point. Due to the large number of Elisa’s products and services, there’s a wide array of relevant combinations.</p><p>In retrospect, if Elisa would not have started to critically reason such service chains for the purpose of the robot back in 2017, we would have struggled to the point of chaos when our service points had to close due to the pandemic.</p><p>However, we didn’t struggle and that’s because we have Annika.</p><p>Annika is a service robot who cares very little about how things have been done until now. In order to assure that Annika would be able to serve each client fully, each product manager must describe every detail and every choice that may come up in the client process. Because of Annika, we have had to create a logic that would prescribe that, for example, a 65-inch TV can only be purchased so that it will be delivered by a courier, independent of which channel is used for the purchase itself. In case of a tablet, on the other hand, the client may choose between various delivery options: courier, parcel terminal, or pick-up.</p><p>This example may seem trivial, but it is in fact essential. A 65-inch TV is not something you will simply sling across your shoulder and carry home in a tram, whereas one can do that with a tablet. However, for the system these two are both simply end user appliances. This means that a large number of Elisa’s current service situations have been described for Annika in a way that focuses on the client experience. Of course it may be argued that building services around the client does not necessarily need a robot, but would we have simply created a spreadsheet for this?</p><p>Anyone who has come to contact with a large service company, knows that is highly unlikely. However, over the course of the past three years, Annika has put us in a situation where we have had to do that and build a completely new service process logic. This has mainly been our product managers’ effort. Either way, as the decision was made on the morning of March 16 that all Elisa service points are to be closed, we only needed a few minutes to identify processes where service points are of critical importance, and to start redesigning these. Yes, we needed just over a day to actually complete the task. That was because we knew exactly what needs to be done and that saved us months of work.</p><p>All our service point employees were able to start work from Tuesday morning onwards side by side with Annika. The pandemic did not lessen the clients’ need for our services and in a situation where a service robot is able to independently satisfy the need of each fifth customer, we needed our service staff to be able to work as well.</p><p>Our staff needed a few days to get used to the new situation, but by today we are working smoothly together with Annika. In the course of this new practice, we have managed to collect highly useful training materials for Annika, that will enable us to learn and improve even more.</p><p>This means that our service staff will be freed of the tasks that are boring for humans and gives them the opportunity to spend more time on clients whose issues need more consideration and human interaction.</p><p>We are grateful to all clients who have been very patient with us at these tremulous times. We also send thanks on behalf of Annika to everyone who has been happy to have her on board. As for other large service companies — we can only recommend automating or robotizing processes — the sooner, the better. No matter whether the implementation of AI is caused by a positive or a negative change (or, indeed, a crisis), it is a decision that will help to gear up considerably and succeed also in the toughest of times.</p><p>Written by <strong>Mailiis Ploomann, Head of Telecom Services at Elisa Eesti AS</strong></p><p>Read more about AI solutions implemented in Elisa from our <a href="https://www.mindtitan.com/ai-customer-experience-telecommunications/">case study</a>.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=8147cff5f0c9" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>