The Ultimate Guide to Using AI in Contact Centers for an Early Adopter

A guide for company leaders to help them understand, how to find and create value for their employees and customers with AI in their contact center.

Date: 19/06/2018
by Yura Filimonov,

Artificial intelligence (AI) has been in the works for decades, but only recently had shown promise for businesses and customers. While there are lots of expectations from AI and some successes, there is also nervousness about how to approach its use in the company.

After reading this guide, you should have a good understanding of what AI is, what value it can bring to a contact center and how to deploy a successful AI project in your company.

1. What is AI and Machine learning

What AI can do

What AI can do now, will be able to do later and won’t be able to do ever.

Source: “What customers really think about AI: a global study”, PEGA

AI is a computer science discipline focused on emulating human intelligence and behavior with machines.

A device with AI is the one, who perceives its environment and takes actions that maximize its chances of success.

AI is usually taught in layers of data, called neural networks, on verified sets of data in a process called machine learning, which also includes AI learning from its own actions. Deep learning is when one uses lots of neural networks to analyze huge amounts of data.

Essentially, AI isn’t a single solution, but a number of ways the real world can be recorded, converted to data, analyzed, categorized and made predictions from by various applications.

AI devices can analyze the surroundings in the following way:

  • make sense of huge amounts of data
  • learn from data and results of its actions
  • make predictions
  • recognize text
  • recognize speech
  • process language
  • recognize emotion from text or speech
  • analyze images and identify objects.

Make sense of data

The main part of building an AI is to feed it tons of data, which it formally describes as objects, relations, concepts and properties, so other software can use them.

Learn from data

Machine learning is the fundamental part of AI, where it can look at data and:

  • find patterns in a stream of input
  • classify objects
  • describe the relationship between input and output
  • predict how output would change, as the input changes
  • reward the agent for good decisions and punish for bad ones.

Make predictions

The inseparable part of AI finding patterns in data is its ability to make predictions, based on that data. That’s why the amount and quality of data are vital to the success of an AI project. A unified data platform is a given for it to happen.

Natural language processing

AI can read and understand texts from huge amounts of data, such as newswire texts. Its immediate applications include information retrieval, text mining, question answering and machine translation, e.g. Google translate.

Humans can use AI’s language processing to speak commands towards applications, receive coherent responses and translate languages on the fly. AI can also analyze the text for specific phrases and gauge client intent.


AI can use various inputs, including cameras, microphones, sonars, etc to make sense of the surrounding world. There are narrower fields of use for AI:

  • computer vision
  • speech recognition
  • facial recognition
  • object recognition.

Computer vision

AI can recognize objects in images or videos, which can be video sequences, views from multiple cameras or multi-dimensional data from a scanner. There are several differences between 2D image processing, 3D analysis of 2D images in computer vision, while machine vision is more about providing imaging-based guidance for automatic inspection.

You might’ve heard about the automatic checkout in the Amazon Go store.

Speech recognition

Also known as speech to text, it allows AI to recognize and translate spoken language into text. It can be used in call dialing, home appliance control, voice search, data entry and preparation of structured documents.

Jon Arnold warns us that enterprise workers should be able to use voice as input in their workplace, since they are accustomed to using it personally.

Facial recognition

As a subset of computer vision, AI can also identify faces from images or video.

When it comes to using it in contact centers, visitor identification in banks comes to mind.

Object recognition

AI from this subset of technology finds and identifies objects in image or video sequence.

Speech and text analysis

Companies can use AI to recognize their customers’ emotions to warn call operators of angry customers, to grade agent work or suggest operators to use soothing phrases or invite a supervisor. Most AI solutions transcript speech to text and analyze vocabulary used, but there are a few that try to decode tones in speech as well.

What isn’t AI

Rules-based software, which can mimic a human expert in a very narrow field, such as an auto-pilot, is not AI, because:

  • it hasn’t been taught on huge amounts of data,
  • it can’t learn from its experience,
  • it can’t predict data from historical data.

Read more about what AI isn’t from Jensen Harris:

You can also read “What AI can and can’t (yet) do for your business” by McKinsey.

Robotic automation

Robotic automation runs on software rules and can be used to automate data entry, moving data from and to legacy applications and other routine actions that don’t require much thinking.

Though it still can be used in a contact center to:

  • automate “cut & paste” operations from one application to another,
  • perform rapid look-ups for specific information,
  • back-office operations, such as processing work cases, keeping track of items, etc,
  • processing & tracking invoices, producing documents, etc.

Nicolas De Kouchkovsky writes that:

“We are led to believe that robots are an application of Artificial Intelligence (AI). It is not the case. If AI can enhance robots, many software robots neither use nor require it. The industry shares some responsibility for that confusion.

AI can and will enhance robots. The two complement each other as described by Nice in its recent announcement of a Robotic Automation Cognitive Framework. But the delineation is important. You can start using Robotic Automation without Artificial Intelligence and its data prerequisites.

You can actually look at Robotic Automation as an enabler of AI. Providers such as Kryon Systems are planning to use it to gather the data needed to enable machine learning.”

2. 5 Solid Reasons to use AI Now

Contact centers right now are in what Jon Arnold calls “the Contact Center Paradox”, where always-on, internet savvy customers expect to get quick, professional help 24/7 as well as a personalized experience. Meeting both of these demands at the same time would have been costly, if it weren’t for the AI, because, given enough data and training, AI can form the base of excellent self-service, competent AI-aided support and personalized experience for every customer.
Those aren’t the only reasons to use AI in a contact center.

  1. CX is #1 differentiator between companies and AI is what can allow companies to continue increasing their CX quality, 24/7, while not exploding from increased costs.
  2. Customers are learning to use virtual assistants, self-service, chatbots and may expect 24/7 customer service. All of these approaches require AI to work well and include some sort of automation.
  3. 91% of companies with top CX and top brands use or plan to use AI, compared to 42% on average. If you don’t, you’ll stay behind.
  4. The earlier you have data, the more data you will have, the better AI you will have.
  5. AI can greatly automate some processes, and reduce costs and increase employee efficiency, as long as you don’t expect to replace employees with AI.

As Jon Arnold says, “Contact centers can’t afford to wait for speech recognition accuracy to reach 100% — if it ever does — to deploy it more boldly. Customer expectations are evolving too quickly to wait for the perfect time, and companies need to be willing to try new things. And certainly the need for contact centers to find a better way is urgent.

The key will be for contact centers and developers to accept some risk to refine these applications, and let these technologies find their legs. Clearly, some customers are ready for this now, and that’s where the focus should be.”

“Twenty-five percent of customer service and support operations will integrate virtual customer assistant (VCA) or chatbot technology across engagement channels by 2020, up from less than two percent in 2017”, according to Gartner, Inc.

“by 2021, Gartner projects that 40% of new enterprise applications implemented by service providers will include AI technologies”

Read more:

3. 8 AI Myths to Beware of

  1. Sophisticated AI can deliver business value without data
    AI needs to analyze lots of data to be able to make decisions or predictions it’s valued for. That’s why it is important to identify sources of data in your organization and to make sure they are open enough to be used in a single piece of software.
  2. Lots of positive data is enough
    Positive data may not be enough to train AI to be unbiased.
  3. Just having data is enough
    It takes a lot of time to collect and prepare the data, because there’s nothing worse, than training AI on lots of bad data.
  4. Companies have access to the data they need
    But that data may be proprietary and the owning organizations may not wish to open it to the public. Other data can be divided between sources and requires agreements between organizations. Or data ownership can be uncertain or contested, making use of AI possible, but difficult.
  5. You need data scientists, machine learning experts and huge budgets to run AI software.
    In truth, you or your developer can use existing AI open source software, Google’s already trained models and pay for the cloud services to create an AI service you can use. It doesn’t take a rocket scientist to connect the AI modules to your existing corporate data as well.
  6. AI will replace humans
    AI is a tool and while some mundane tasks can be automated even in highly-skilled professions, the most value comes from analyzing lots of business data to help people make better decisions faster. Humans are still needed for empathy, persuasion or the personal connection people appreciate.
  7. “By one estimate, a human brain has more switches than all of the computers, routers, and internet connections on Earth. So it’s not really surprising that the technology available now cannot duplicate human thought.”

  8. AI software is autonomous
    While AI can automate some tasks, AI does require humans to:

    • determine, where to use AI
    • prepare data
    • write the algorithms
    • train the AI
    • interpret and determine the value of the results.
  9. AI is unbiased
    The AI model is still the result of the data it has access to. If it has been trained on biased, insufficient data or trained erroneously by a human, then it’ll be biased as well.
  10. Read more:

    4. Where You Can Use AI in Contact Centers

    Mary McKenna, Director of Product Management, estimates that 30 to 50 percent of human call center tasks can be streamlined with AI technology.

    “LivePerson, a customer service platform provider, in a recent interview reports up to 35 percent efficiency gains with “AI-assisted human agent” model and DigitalGenius’s CEO Dmitry Aksenov shared that as of December 2016, 30 percent of KLM cases are resolved with the power of AI, and that percentage is on the rise.”

    “How artificial intelligence is transforming enterprise customer service”, Adelyn Zhou, Forbes.

    Computers can analyze huge amounts of data and dig into it better, than humans, which means they are great at solving specific problems, if they have enough data. It doesn’t mean AI will replace humans, since AI still can’t do creative work adequately.

    Since enterprises and contact centers are awash with data, it makes sense to use AI to augment human work, bringing the best worlds together.


    “DMG-research found that self-service is the channel of choice for all generations, as long as the solutions work well. However, when an issue is time-sensitive, emotionally-charged or complex, customers are going to call, as they are looking for the human touch.”

    “AI and Automation is the Path to the Future for Contact Centers”, DMG Research

    Dumping all the contact center work on self-service is what everyone thinks, when hears “AI in a contact center”, but in reality its quality isn’t good enough to do all the work.

    “Existing self-service capabilities are limited and often ineffective. There is a clear need for better forms of automation, and this is where the promise of AI really resonates”
    “AI in the contact center”, Jon Arnold

    Though current self-service AI tools aren’t perfect, they can:

    • authenticate customers
    • process payments
    • provide the simplest info
    • deliver personalized knowledge base information.

    This can be done via an intelligent IVR, but also through a chatbot.


    Example of chatbot on mobile

    Image credit: John Jackson

    A chatbot is an application, which customers can get automatic responses by using text or voice queries, in contrast with regular live text & video chats, which require contact center agents.

    The simplest chatbots can only provide a narrow field of information, while chatbots with natural language processing can participate in meaningful conversations, have a sense of humor or even a distinctive snarky character.

    Since it’s too early to expect AI to automate everything, it’s a good idea to pick the chatbot that’ll automate the simplest self-service queries and send the customers to support with more complex issues.

    When deciding to use chatbot, you need to:

    • know, which customer problem you are trying to solve, and how it should be solved: automated or together with agents
    • decide, whether you need a sophisticated AI solution or a simple rule-based chat-bot
    • use a chatbot to both automate and enhance agent work
    • estimate, if your contact center solution can handle the chosen chatbot.

    To learn more about how to get started with a chatbot in a contact center, read this Cisco’s whitepaper, authored by Jon Arnold, “AI in contact center and chatbot strategies” (PDF).

    However, don’t be charmed by the simplicity of chatbots: you still need to focus on the data and how you use it, rather than the customer interface.

    Then again, as Shep Hyken says, chatbot does offer exciting benefits:

    • 24/7 support,
    • instant reaction,
    • personalized experience,
    • can build relationships with customers by proactively contacting them.

    Of course, only the recent scandal about Google Duplex brought the issue of non-disclosing AI bots to the public, which means that there will be guidelines for informing customers, whether they are talking to a bot or a human.

    Examples of chatbots for contact centers

    “The WeChat Messenger bot deployed by China Merchant Bank – one of the largest credit card issuers in China – is another example of a front end bot. According to the AI technology provider Xiaoi, the China Merchant Bank’s front-end bot handles 1.5 to 2 million customer conversations per day , an inquiry volume that would typically require thousands of additional employees to answer. As most questions relate to card balances and payments, automation via a bot interface presents a relatively easy and cost efficient solution.”

    “How Artificial Intelligence is Transforming Enterprise Customer Service“, Adelyn Zhou, Forbes.

    “Using the Artificial Solutions platform, a major energy producer and supplier of oil and gas began rolling out an AI-based chat-bot for its Web chat channel in January 2016. The company started getting measurable results within two months, and by November of 2016 achieved the following:

    • 43% reduction in telephone call volume to human agents
    • Real-time handling of research requests that previously took hours or days
    • 97.4% of customer questions correctly understood by the virtual cognitive agent (VCA)
    • 74% of issues successfully resolved — and contained — by the VCA
    • 98.8% of VCA answers met or exceeded customer expectation”

    “Case study: customer engagement with Virtual Cognitive Agents”, Chris Vitek, NoJitter

    Ocado customers can email, tweet or call Ocado -they don’t need to fill in any forms or self-categorise their emails. Instead, all messages gets delivered into a centralised mailbox no matter what they contain.

    Ocado’s AI software uses NLP to parse through the body of the email and creates tags that help contact centre workers determine the priority of each email, depending on its topic and customer’s level of satisfaction – ironically and pragmatically, happy customers get less priority.”

    American Automobile Association has used SmartAction omnichannel chat bot to:

    • provide assistance over the phone during peak hours due to bad weather, holidays, etc
    • identifies the member and ensures their safety
    • locates the member via an SMS and deploys the tow truck driver.

    AAA has got pretty good results:

    • SmartAction handles 200k calls every month without any hold times in peak volumes
    • decreased cost-per-call by 50%
    • achieved the same automation scores in CSAT surveys as with the live agents: 4.5 of 5.

    Internet of things

    As with other technologies, IoT devices can be used in concert with a contact center and usually have lots of sensors to collect data to send it to the datacenter for analysis. The huge amount of data, both from a single person and thousands of different customers or devices, allows AI to make reliable predictions about future needs of customers and devices themselves.

    For example, a broken fridge could:

    • identify a problem
    • alert the owner of a problem
    • order a replacement part
    • send a description of a problem to the contact center, when the customer calls it.

    Michael Ringman, CIO of TELUS International, tells us that:

    “The contact center will play a critical role in delivering a great customer experience with the complexity of IoT-interactions requiring more support from customer service agents, not less.”

    The contact center leaders will need to hire not only friendly, but also tech-savvy agents with problem-solving skills, who will be able to help customers with IoT devices in the context of their environment. To help that, there need to be extensive knowledge bases across a large number of interacting IoT devices and, perhaps, blended AI to help the agents to sort through the huge amount of data.

    Another example of smart things is that most car drivers with Progressive Snapshot can get insurance discount around $130, if they drive safely. The app can also advise them, how to drive safer, and get a bigger discount.

    “Over time all problems will become IoT problems”, meaning that everything will have a sensor, including things embedded under our skins.

    Tom Siebel, C3

    Conversational IVR

    AI can work not only with data, but also with speech – which companies can use in IVR to create a natural conversation to help solve the customer issue, perhaps without even contacting an agent. The use of AI can also help with first call resolution, because AI can not only understand the question, but route the customer to the appropriate solution quickly.

    Examples of AI IVR

    Interactions seek to take IVR to a new level and break out of the mold of inflexible, difficult to navigate IVR trees. The goal is to create a completely natural interaction between customer and software, such that, in the best case, the customer doesn’t even realize the interaction has been handled by a machine.

    The AI IVR can also route calls for higher ROI, quick resolution or to empower customers to rely on self-service.

    Digibank is the first digital-only bank in India. It operates with no branches and only one-fifth of the resources of a traditional bank. Digibank assistant was central to its success: it handles over 80% of customer inquiries with only 20% requiring chatting with an agent.

    Royal Bank of Canada’s new Credit Card Service Line (PDF) reached impressive precision levels, 70% for authentication and 93% for understanding. It not only drove up self-service usage but improved customer satisfaction.

    Perficient made an IVR solution with IBM Watson for a large health insurance provider and it delivered many benefits:

    • Drastic reduction in requests for a live agent, leading to substantial cost savings
    • Average call times dropped from eight minutes to three, highlighting the effectiveness of the cognitive agent
    • Enhanced logging capabilities allow for better call metrics and analysis, since Watson can record and report on all conversations.

    The Convergys Conversational platform eliminates the need for an Interactive Voice Response (IVR) menu. Companies receive and respond to user questions made over virtual personal assistant (VPA) speakers as users are instantly linked to the entities desired.

    FedEx updated their IVR with an NLP-powered IVR Nuance and achieved the following results:

    • eliminated 11k phone calls,
    • increased automation by 6% (+3% over their goal),
    • saved money, because customers preferred to use this kind of self-service.

    Read 3 other IVR Nuance case studies.

    Intelligent Call Routing

    One of the ways AI can improve contact center is to personalize customer journeys and experience by analyzing historical data in CRM and live data, and adapting CX on the fly. An intelligent solution should be able to direct the customer from any channel, give him a chance to self-service or to send him or her off to the most suitable agent.

    Predictive routing software uses historical and real-time data, along with AI. It continuously and automatically uncovers meaningful factors that influence the outcome of interactions between customers and employees, such as:

    • favored communication channel,
    • product(s) purchased,
    • past service requests,
    • recent transaction activities,
    • employee’s tenure,
    • employee knowledge, skills & interaction history,
    • and business outcome data

    to predict the optimal customer-employee match for the desired result.

    Examples of intelligent routing

    Interactions provides relatively simple routing. But, for others, it offers a self-service solution that truly takes advantage of its adaptive understanding technology to deliver answers, reduce channel hops and handle time, and increase satisfaction.

    In many cases, the solution helps identify revenue generation opportunities and routes them to the appropriate channels. In others, simple self-service scenarios result in information being quickly delivered to customers.

    Interactions’ marketing communications manager Dan Fox says, “The real ROI, though, comes from situations where Interactions is able to show customers how to leverage its virtual agents to create self-service opportunities out of traditional agent-enabled interactions.”

    Using technology like IBM’s Watson Conversation, Watson Discovery Services and Salesforce’s Einstein services, customers will be able to first seek an answer via live conversation with AI. If the AI bot believes the conversation is too complex, the entire conversation can be routed to the best live agent to support the question.

    “Leading Canadian communications and media company Rogers Communications, Inc. increased retention by nearly 3 percent and reduced average handle time by 7 percent with Genesys Predictive Routing.”

    “Predictive Routing is a practical and powerful way to use machine learning to align customer intent with the best equipped agent. We see great potential in using the Genesys solution to deliver smarter and faster service that respects our customers’ time, sets our employees up for success, and produces stellar outcomes for our business.”
    Kevin Jolliffe, vice president of Enterprise Planning for Rogers

    “A mobile telecommunications operator in Western Europe increased its Net Promoter Score® (NPS®) by four points, improved FCR by nearly 4 percent, and reduced average handle time by almost 3 percent. In addition, an Australian telecommunications and media company achieved a sustained, all-time high NPS.”

    “New Artificial Intelligence-Powered Routing Engine by Genesys Creates World’s Most Predictive Customer Experiences”, Genesys

    Agent Assistance with Blended AI

    “AI isn’t ready to automate CX completely right now. I like companies, who are using machine learning and big data to make agents smarter before AI. I think it’s an AI “co-pilot” situation with live reps at the moment – it is also a good idea to let the reps train the AI in real time.”

    Daniel Dougherty, Managing Partner at Dougherty Martinsen

    Blended AI is when AI software is used to analyze business data and make suggestions to agents in real time, so they could provide more personalized, timely and accurate customer service.

    Since only a part of human tasks can be seamlessly automated with or without AI, AI can still be applied to data in the rest of the company – and that’s where it needs human empathy, just as much as humans need AI’s capacity to analyze huge amounts of data, categorize data, point out outliers and make predictions.

    Forrester's research shows that blended AI is best

    Forrester’s research says that AI can handle the simplest requests and decrease costs with automated conversations, but it needs human’s touch to handle complex issues.

    Source: “Artificial intelligence with the human touch”, Forrester for Genesys

    For example, Forrester in its study “Artificial intelligence with the human touch” says that AI can’t beat humans in terms of understanding emotions and building trust and that using both humans and AI for a task improves customer and agent satisfaction.


    That’s why blended AI is a major part of using AI in contact centers and it will remain so in the foreseeable future.

    Jon Arnold suggests that “a strategic use case would be developing a chatbot that can comfortably handle a specific set of inquiries, but can then seamlessly extend the session to a live agent when human interaction is needed.”

    Brad Shimmin said of Kate:

    “There’s no reason for a company like Genesys to try to solve every domain specific problem on its own, so the company invites its partners to bring in the most appropriate chatbot, maybe those built on Watson, Einstein, or even Dr. AI.

    I think this blended approach will become the norm over time, given the open nature of most AI libraries and the public cloud platforms where they reside.”

    Examples of Blended AI in Contact Centers

    Kate is described as combining artificial intelligence, bots, machine learning, and micro-applications so companies can deliver personalized, proactive, and predictive experiences while running a smart business. The combined power of Kate and live staff to solve customer problems is what Genesys is calling “blended AI.”

    The differentiating attribute of Kate’s brand of AI — that she’s here to work with and assist live agents, not replace them.

    Another example comes from a Swedish bank with Amelia, a virtual assistant:

    “Amelia has an understanding of the semantics of language and can learn to solve business process queries like a human. It can read 300 pages in 30 seconds and learn through experience by observing the interactions between human agents and customers. If Amelia can’t answer a question, it passes the query on to a human, but remains in the conversation to learn how to solve similar issues in future.

    The Swedish  bank is confident the software will prove successful: “Within the first three weeks, it held over 4,000 conversations with 700 users and was able to resolve the majority of those queries independently, allowing employees to get consistent support without delay.”

    “To manage an increasing volume of queries, while increasing customer satisfaction, TravelBird looked to new computer technology to take charge of repetitive tasks like tagging and macro suggestion based on historical data and continuous AI learning.

    By adding DigitalGenius to their Zendesk agent interface, TravelBird has been able to better manage incoming customer enquiries. After only 3 months, the AI is able to cover more than 65% of the incoming queries. Pre-filling accuracy by the AI has reached 95% and agent macro usage has increased 200%. Thanks to DigitalGenius, agents have saved more than 40,000 clicks since the deployment and they feel very positive about AI helping them with repetitive tasks.”

    StarOfService looked into AI to help with what it does best: learn from historical data, classify incoming enquiries automatically, suggest the best answer to incoming questions and automate responses for the most common questions, in order for the StarOfService team to cope with the higher number of enquiries.

    By implementing DigitalGenius into their Zendesk agent interface, StarOfService has been able to significantly reduce the average time spent per case. After only 3 months, the AI is able to correctly predict 97% of the cases, average handling time went from 6.1 to 3.3 minutes (46% AHT reduction) and more than 250,000 clicks have already been saved.”

    “One of the ways AI has been used is to monitor and analyze speech patterns and inflections of callers, as well as reviewing specific words, to determine when an interaction may be in danger of escalating. Indeed, “sentiment analysis,” where a system will detect changes in tone, speech patterns, or volume, can often be useful not only in addressing a situation in real time that needs to be escalated to a manager, it can also be used as a training aid so agents can learn to better recognize signs of stress or anger during an interaction. It can also suggest ways for an agent to reduce the stress level of a conversation.”

    “AI in contact centers”, Keith Kirkpatrick, ACM

    Customer voice recognition

    An interesting use of voice recognition is to recognize calling customers, when they are calling the contact center.

    For example, Nuance has a voice recognition software suite, where customers either use a specific voice-phrase as a password, or it can simply identify the customer from their call.

    Barclays bank uses a similar technology to identify customers by voice and it favors those, who use phone banking regularly, rather than visit the branch or use the website/app.

    Brett Beranek, director of product strategy, voice biometrics, Nuance Communications said:

    “Following the deployment of Nuance’s text independent voice biometrics solution by Barclays Wealth & Investment Management in 2013, 93 percent of customers rated Barclays at least 9 of 10 for the speed, ease of use and security of the new authentication system.”

    Customers rarely call banks

    50% of customers called a bank once in the last 8 months.

    However, HSBC had a misidentification, when their client’s brother pronounced the preset phrase to be identified as the bank’s client. The bank said they increased software sensitivity after the incident and that:“Twins do have a similar voiceprint, but the introduction of this technology has seen a significant reduction in fraud, and has proven to be more secure than Pins, passwords and memorable phrases.”Another drawback of this technology is that aging customers, who call the bank very rarely, are likely to have trouble with being recognized, because voice changes with age, within as little as 2 years.

    For example, if you call the bank less, than 3 times in 8 months, you have more than 10% chances of not being recognized.

    “But the research is worth considering as banks continue to deploy voice recognition to authenticate callers to their call centers. USAA, Wells Fargo, Eastern Bank, Tangerine, Barclays and HSBC are among those that use the technology.”

    “Voice recognition’s surprise pitfall: aging customers”, Penny Crosman, American Banker.

    In Singapore 56 percent of firms indicated that they have either implemented voice recognition, or are expanding or upgrading their technologies, compared to 46% in the US and 33% in the UK.

    You can get free Forrester, APAC, etc research on customer identification here.

    Virtual assistants

    Virtual assistants are AI-powered audio devices that accept voice commands, requests and queries, and collaborate with the user. Usually, they help customers in their homes, but contact centers can use them to assist the agents: an always-listening virtual assistant that can proactively contribute to conversations.

    Virtual contact center assistants make it fast and efficient for customers to obtain the help they need. They minimize long hold times and offer digital verbal interactions that can answer their questions, identify sales opportunities, and solve problems.

    SkySwitch for Alexa ties AI to voice, where end users can use Amazon Alexa to voice-enable the calling process on their PBX. Instead of going to the phone and punching in the extension, you just dictate the request to Alexa. Now think about someone who makes 30-40 calls a day, and this application doesn’t sound so trivial.


    “The hockey stick growth curves here tell a strong story, both for the installed base of Echo endpoints, and the applications ecosystem – which Amazon calls “skills” – that is exploding to drive adoption. As with anything else, value is defined by utility, and once workers get comfortable using Echo to dial their calls, it won’t take long for other AI-driven productivity applications to emerge.”

    “Reimagining Voice in the Age of AI”, Jon Arnold

    Cisco Spark in VR, which is now available in the Oculus store. The AI assistant within Cisco Spark in VR can do things like build a customized virtual meeting space and populate the room with 3D images. But it can also learn and bring in knowledge and resources. In a demo, the team prompted the Spark Assistant to learn about jet engines and after a quick scan of Wikipedia, it was able to answer questions like how much a jet engine weighs or what causes jet engine noise.

    Learn more about Cisco and AI.

    Jupiter Next hotel replaced the guest phones with the Roxy virtual assistant to save on utility and service costs, while adding convenience to the customers. Roxy has a digital display for those, who are not comfortable with commanding Roxy by voice.

    Phonesuit, a hotel voice technology provider, started using Roxy as well.

    Predictive analytics

    How predictive analytics works

    How predictive analytics works.

    Source: “What is predictive analytics?”


    Predictive analytics is used, when AI analyses existing data, monitors real-time data and makes predictions on what might happen.

    In contact centers, software can aggregate available data about a customer and compare it to predictive models for behavior and emotion. By analyzing the data, the software can inform agents of customers’ needs and preferences, enabling them to provide a higher quality customer interaction.

    For example, companies can use predictive analytics to:

    • identify less satisfied customers and make them personalized offers to entice them to interact with the company
    • anticipate phases of customer lifecycle and provide help or offers, when customers need them, but haven’t yet contacted the company
    • identify dissatisfied employees that are likely to quit and work on retaining them
    • get better hires for the contact centers by finding the traits of successful employees and looking for them in candidates
    • learn, when to expect more calls, based on a new major complaint from customers

    Examples of predictive analytics for contact centers

    The combination of Genesys and Altocloud will enable businesses to score and predict a consumer’s journey in real time across channels, while they are live on a website, using a mobile app or in a conversation with an employee. This is achieved through AI and machine learning, which use pre-defined personas and past behaviour analysis to automatically predict consumer outcomes and give context to the customer’s journey. As a result, organisations can deliver the next-best action by the right employee for improved success rates.

    Mattersight’s solution uses algorithms to identify the best agent to handle each caller based on the agent’s past performance and personal strengths, and the customer’s personality and other behavioral characteristics. Also, it combines existing call center data with new interaction insights to assign a predictive score to each call. By capturing, analyzing, and decoding every second of the customer-agent conversation, Mattersight can automatically predict CSAT, NPS, customer effort, and churn.

    “By applying Mattersight Predictive Analytics to the call center, our customers have been able to:

    • Drastically reduce the need for (often inaccurate) customer surveys
    • Design proactive response strategies for at-risk customers
    • Improve business processes that create the most customer friction
    • Decrease customer attrition by as much as 33%”

    “A company called CornerstoneDemand has come up with a data-based ‘Evolve’ solution for this, putting applicants through an automated application process, including recording their voice and giving them simulated calls to answer. The company then uses data analytics to predict who should get an interview, who should be interviewed with an eye toward certain potential problems, and who shouldn’t get an interview.”

    “How analytics, Big Data and AI are changing call centers forever”, Bernard Marr, Forbes

    Via Cornerstone Selection, candidates are given a preview of job responsibilities to determine skill and interest level based on real-life tasks. They are also evaluated on the personality traits known to correlate with productive call center performance: engagement, high levels of customer service and lower call times. Candidates are then given skills tests to ensure their ability to be successful in a call center environment.

    Overall, Cornerstone HR hiring suite with AI analytics seems to help call centers:

    • reduce attrition by 20-30%
    • increase employee productivity by about 10-15%.

    AnswerOn had used predictive analytics to identify agents likely to leave and pointed out to call center managers, why the agents might leave, and suggesting a specific coaching program. For a call center of 5000 agents, AnswerOn achieved the following results:

    • Average reduction in agent attrition: 30% over two years
    • Average reduction in churn rate for targeted agents: 65%
    • Reduction in training costs: 20%
    • Program margin improvements: 10%

    Speech analytics

    “Speech analytics and placing machine learning on that is the next step, imho.”

    Daniel Dougherty, Managing Partner at Dougherty Martinsen

    “Let’s look at a concrete example of analyzing the customer sentiment in a contact center. Almost all the inbound calls to the contact center are recorded for random sampling. A supervisor routinely listens to the calls to assess the quality and the overall satisfaction level of customers. But this analysis is done only on a small subset of all the calls received by the call center.

    This use case is an excellent candidate for AI APIs. Each recorded call can be first converted into text, which is then sent to a sentiment analysis API, which will ultimately return a score that directly represents the customer satisfaction level. The best thing is that the process only takes a few minutes for analyzing each call, which means that the supervisor now has visibility into the quality of all the calls in near real-time. This approach enables the company to quickly escalate incidents to tackle unhappy customers and rude call center agents.”

    “3 Steps To Embedding Artificial Intelligence In Enterprise Applications”, Janakiram MSV, Forbes

    Examples solutions for call analysis

    Cogito has developed a real-time conversation-analysis tool based on behavioral science and deep learning. Their AI listens to conversations for both content and tone. They claim it can detect mimicking, change in volume, change in pitch, etc. to gain real-time insight into how customers are feeling and how all company calls are going. It provides real-time suggestions to customer service representatives to improve the call and evaluate performance.

    One of the first big tests of their system was at insurance giant Humana. During a six month trial involving 200 agents, calls that used their system resulted in a 28 percent improvement in Net Promoter Scores, a 6 percent improvement in issue resolution, and fewer callers asking to speak to a manager.

    “Use Cases of AI for Customer Service – What’s Working Now”, Jon Walker, TechEmergence

    Transcosmos uses voice recognition, so its “clients can use uniform criteria to “auto-evaluate all calls”, including the evaluation of basic support services such as “greeting”, “use of technical terminology” and “timely response”, which traditional “human” based monitoring systems could not deliver. Based on the monitoring results, transcosmos will add new KPIs such as stability of service and KPI achievement ratio.”

    Performance analysis

    With the huge amounts of data have, it takes algorithms to analyze them. To analyze that data and to receive analysis with actionable results in real time, companies use AI in contact centers.

    Мachine learning applies to a call center, customer care team in multiple dimensions: risk estimation and insights, performance/KPI analytics, relationship networks and correlation impact between various performance indicators in your company.

    AI can also be used in gamification to reduce work attrition, improve KPIs and improve employee education.

    RapportRoadmap takes live chat messages and their outcomes and analyzes, which parameters of the conversion influenced the sales and by how much. It looks at how keywords, message format, length, frequency, punctuation and capitalization, and emotionally intelligent conversations influence customer engagement, retention, and conversion and makes corresponding recommendations to the contact center directors.

    They then can use RapportCoach and the KPI data to coach the agents how to adapt their conversations to increase conversions, order size, etc.

    Read more:

    5. Things to Remember Before Using AI

    “Each and every project becomes faster. The upfront costs, the nonrecurring costs, of development are lower. And we’re able to, with each project, add more value and more business content to that data lake.”
    Matthew Evans, vice president of digital transformation of Airbus

    Before jumping on the AI bandwagon, you need to remember a few things.

    1. The AI is as good as the data you train it on.
    2. AI practically requires a unified data platform, including both internal and external data sources.
    3. While AI can automate processes, it’s more efficient to have AI help agents, not in the least, because then it’d get their support
    4. The more context your AI has, the more personalized the CX is.
    5. Even if you only start using AI as a self-service chatbot, the AI project is still an important one, because it can be viewed as a starting point of using AI for the entire customer journey
    6. AI constantly learns on new data, so it is different from the static state of the usual software.
    7. The more data your AI learns from, the more valuable it becomes not just for one project, but for your other AI projects as well.

    6. How to start using AI in your contact center


    “The leaders not only have a much deeper appreciation about what’s required to produce AI than laggards, they are also more likely to have senior leadership support and have developed a business case for AI initiatives.”
    the MIT Sloan Management Review

    When it comes to choosing your first AI project for a contact center, Jon Arnold’s opinion is that:

    • you should focus on a very simple case, text-based, probably a chat-bot
    • immerse yourself into the AI environment with your first project without ROI being a priority,
    • yet do track and analyze how well your AI project is accepted and how well it helps customers.

    “Creating an AI project for the contact center means defining the business problem AI should solve and finding the right developer to make the project a reality.”

    Jon Arnold, “How do you get started with contact-center AI?”, TechTarget

    Though setting up a chatbot to test AI seems like a simple thing, there are certain obstacles and some things can go wrong. Here’s what you need to do, if you want to build an AI app in your company:

    1. Have a good understanding of what AI can do and how it works.
    2. Determine, which problem you are trying to solve.
    3. Look at both customer-facing and internal problems.
    4. Use technological advances: gathering & processing a new source of data and making autonomous decisions.
    5. Find a business case for AI.
    6. Don’t yet define ROI, but focus on a proof of concept.
    7. Get a buy-in from the trenches, someone from the contact center or IT, to make it relevant.
    8. Get executive support and make sure the executives understand AI.
    9. Determine, where and how you’ll get the tons of data that’s needed to teach AI.
    10. Determine, which type of vendor you need: a hands off one or one that’d help you integrate the solution.

      “I think that to wait for the data would be a mistake. I am also a strong believer in business that if you wait for the right time then you fall too far behind. It is always the right time to innovate. So I would start with making sure my digital strategy was building in a holistic approach for AI. Then I would start to implement the lower hanging fruit. Things like agent assistance in chat-bots and the rpa stuff like process automatons on the back end. It allows you to start the AI journey and collecting the appropriate data. This allows a company to start down the path and find the right partner to work with.“
      Fred Stacey,

    11. Acquaint yourself with available AI solutions.
    12. Pick the one that matches your use case and level of automation.
    13. Make sure the solution is integrated into your contact center and CX processes.
    14. Ensure high voice quality for AI to work with, which will otherwise result in bad data.
    15. Ensure the data you work with and create is secure.
    16. Train the AI on your business-specific data.
    17. Test extensively to avoid customer dissatisfaction and public failures.
    18. Attract or develop the right AI talent.
    19. Build an AI team.
    20. Organize employee communication, education, and re-training on the topic to reduce employee resistance.

    Read more:

    7. How to implement AI yourself

    If you want to create an AI software yourself, you’ll need to do the above and:

    1. Identify sources and have access to tons of data to teach the AI.
    2. Develop good practices for working with data.
    3. Integrate all the data points.
    4. Decompose processes to see, where AI can be used.
    5. Start small in text, such as an SMS bot or a chatbot, because it won’t use as much resources.
    6. Choose the pilot to deliver the most value with less resources.
    7. Determine integration, capabilities and scale requirements for a fully operational AI process.
    8. Use existing AI APIs in your application or train your own machine learning models by using Google’s AutoML.
    9. Allow enough time for training.
    10. Use explainable AI to make AI’s decisions transparent. Use complex algorithms or deep-learning cautiously: such AI is capable of making intuitive creative decisions, which are hard to track to the origin.
    11. Start new processes between developers and product managers.
    12. Measure agent and end-user experience, analyze, improve.
    13. Move on to more complex projects that involve speech, multiple APIs and processes.
    14. Consider risks and be ready for them.

    For companies with limited AI experience it’s reasonable to work on a use case that is likely to deliver value, is reasonably well defined, and is only moderately complex. This test will help the organization gain familiarity with AI and will highlight data or data integration needs and organizational and capability hurdles—critical inputs for the next step.

    Susan Etlinger reminds us that AI needs to be customer-centric and that it must:

    1. be clear, useful, and satisfying (even delightful) for the user.
    2. understand and respect people’s explicit and implicit needs.
    3. be transparent, secure, and act consistently.
    4. be free of bias that could cause harm—in the digital or physical world, or  both—to people and/or the organization.
    5. have clear escalation and governance processes and offers recourse, if customers  are unsatisfied.

    As Andrew Ng says, a successful AI company has:

    • strategic data acquisition
    • unified data warehouse
    • new job descriptions
    • an AI team.

    “The center of excellence is not intended to be the group that will provide all analytics for the entire organization. It provides expertise, guidance, and direction to other internal teams that are working to deploy AI and analytics,” says J.D. Elliott, director, enterprise data management, TIAA.

    They should look how to integrate AI within the company and look for use cases.

    There can be an inter-department AI team with a product manager, a developer and a developer with specific business domain knowledge to bring expertise to departments and ensure smooth integration, when they start their own AI projects.

    “CIOs can give up AI project ownership to individual teams, but can be responsible for how AI augments rather than replaces human work and convinces CEO and the executive board of AI’s value.”

    Zeus Kerravala, founder and principal analyst at ZK Research

    Worth reading:


    8. How to overcome obstacles to AI adoption

    The question of how fast companies are in adopting emerging technology raises its head again:

    “Three-quarters of executives believe AI will enable their companies to move into new businesses. Almost 85% believe AI will allow their companies to obtain or sustain a competitive advantage. But only about one in five companies has incorporated AI in some offerings or processes. Less than 39% of all companies have an AI strategy in place. The largest companies — those with at least 100,000 employees — are the most likely to have an AI strategy, but only half have one.”
    The SloanMIT Review: “Reshaping Business with AI”.

    Sloan survey respondents rank business issues, such as finding a business case for AI, higher, than technological problems.

    This data corresponds with from other similar surveys.

    AI adoption

    More than half respondents are researching or planning to use AI, while about third are piloting/implementing.

    Source: Gartner Research Circle

    “Organizations are still gathering information to inform their AI adoption strategy to facilitate decision making and automate processes. As a result, an AI initiative is most commonly deemed successful when it improves decision making and process efficiency.”

    Whit Andrews, Gartner.

    As Sloan has built a composite index of organizational understanding of AI, based on the survey questions related to AI understanding, combined with the level of organizational adoption of AI. They split the organizations into four types: pioneers, investigators, experimenters and passives.

    Companies at different level of AI adoption face different challenges:

    Let’s see, how companies can overcome issues of adopting AI.

    Understand AI

    While the execs don’t need to dive into the technical nuances of building AI models, they need to understand how AI learns from data, what AI can do, where it can be applied and that AI can suggest decisions, better than gut instinct.

    Recognize value in AI

    The best way to demonstrate value in AI is to find a problem in your company that can be solved by AI. Also, reading AI case studies should help.

    “The whole time with big data and AI combined the company can recognize more buying patterns and improve the customers next visit. The end goal and true ROI is from gathering all of the data points to make the customer experience as seamless and effortless as possible no matter the channel the customer is using. Then allowing the AI to handle what it can in combination with the agent that is best suited to resolve the issue. It can empower the agent and the customer and restricting it to an individual channel would be a mistake.”

    Fred Stacey of

    “Companies looking to achieve a competitive edge through AI need to work through the implications of machines that can learn, conduct human interactions, and engage in other high-level functions—at unmatched scale and speed.”
    “Competing in the age of AI”, BSG

    Organize for AI

    While 70% of the passive companies haven’t specified responsibilities for AI initiatives, the pioneers have a mix of centralized, decentralized and hybrid organizational structures, though, even 30% of them haven’t put anyone in charge of AI projects.

    A hybrid organizational model seems to be most logical for the job, where the main expert group will provide expertise, guidance and direction for internal teams, who deploy AI.

    Jessica Tan of Ping An, who employs 110 data scientists, said that the main challenges were:

    • getting units to work together
    • acknowledging the fact that “humans don’t want to train algorithms”
    • establishing centralized and decentralized technology teams
    • finding the right people.

    They are looking for 3 types of people:

    • technical people, who have the means to try new ways of working
    • technical people, who understand business domains
    • consultants or project managers, who can bring them together.

    Have an AI strategy

    “Larger companies are more likely to have an AI strategy in place, but only half have it.”

    Reshaping business with artificial intelligence”, MIT Sloan Management Review.

    Yet again, this figure shows us how companies are different in terms of adopting AI strategy.

    According to Amy Hoe, chief technology and operations officer of insurer FWD Group, access to data as key for competitive advantage for her company. FWD aims to secure a wide range of data sources, including partnerships with other companies, such as:

    • telecommunications companies
    • ride-hailing services
    • its customer base
    • agencies
    • social media
    • the public domain
    • external data analysis providers.

    “Organizations that flourish with AI will become proficient at acquiring data strategically. They won’t merely have data. They will understand where and how to get more.”

    “How to create a successful AI strategy”, Amit Ashwini

    Focus on the data

    Since AI runs on data, having a lot of it and of good quality is paramount. Since AI algorithms are fairly easy to find and you can already find trained models, the differentiator in AI quality will be access to data and its quality.

    “For AI to be effective within an organization, CDOs must help establish a data-driven culture – information as a second language, if you will”

    “Artificial intelligence at the enterprise”, Mike Rollings, research vice president at Gartner

    To see an example of an AI intentionally trained on bad data, read this article.

    Find a business case

    “Well, strictly speaking, we don’t invest in AI. … We’re always investing in a business problem.”

    Matthew Evans, vice president, digital transformation, Airbus

    It makes sense to:

    • figure out, which technologies and data AI uses,
    • split your business processes into segments and see, where big data can be analyzed to introduce an insight about customers to help employees,
    • prioritize them by possible value per investment cost.

    “One sensible approach to setting priorities is to create a heat map of RPA and AI opportunities across relevant products and processes, plotting value created against time to implementation.”

    “Powering the Service Economy with RPA and AI”, BCG

    Gartner’s survey demonstrates that most companies start with AI solutions that improve decision making and process automation. But as organizations begin to consider more symbiotic uses of AI, they will be less likely to simply replicate the steps that a human performs to reach a particular judgment, and will instead use the relative strengths and weaknesses of both to maximize value generation.

    Find ways to be higher in investment priorities

    By finding a lucrative use case in the company and planning a pilot, where you don’t need to hire data scientists, you can demonstrate value and move up the investment priorities.

    Even on their own, companies have a chance to use at least two dimensions, natural language interfaces and data-driven insights, to obtain noticeable value from AI.

    Get executive support

    While some projects can start on the grassroot level in separate departments, to introduce AI to the enterprise level C-level support is needed. Here educating the execs on what AI does, how it works in general and how it can be used in the company should help them embrace the powerful concepts.

    Then it comes to translating the results of AI software into money.

    “If the IT leader can explain it in those terms — that companies that have used it have seen customer experience with a net promoter score grow by “X” percent, that’s a great way to justify that purchase versus having something that maybe drives cost down — which is interesting, but isn’t really going to change the business.”

    Zeus Kerravala, founder and principal analyst at ZK Research

    Solve technological issues

    It’d be naive to expect to run AI without a unified data platform, because it needs to access lots of data to make predictions, which can be shown to a multitude of employees. Essentially, it means you need to go digital and omnichannel before you start using AI in your contact center.

    Help employees embrace AI

    “… many people resist AI because of the hype surrounding it, its lack of transparency, their fear of losing control over their work, and the way it disrupts familiar work patterns.”

    Brad Power, “How to Get Employees to Stop Worrying and Love AI”, HBR

    That’s why it is important to build AI solutions that augment human actions and demonstrate the value of adding AI to the workplace.

    “Some employees might think that companies can use AI to spy on them or their customers,

    the Big Brother thing. You may want to ensure they trust AI before you go too far down the road. So it helps to make it transparent:

    • where and how AI is used,
    • why AI makes particular decisions

    to show that AI is used for good.”

    Jon Arnold

    The best way to avoid the “AI vs workers” mindset is to make sure AI augments what humans are already doing.

    “Helping agents understand how contact center analytics can contribute to their success is key. … Speech, text and desktop analytics can empower agents and transform their work experience. Here are a few of the ways agents will benefit from these solutions:

    • real-time agent guidance or next-best action,
    • analytics quality assurance,
    • identify compliance issues,
    • an early warning system.”

    “How do we minimize the perception of big brother?”, Donna Fluss

    Establish a clear plan for AI projects, including where AI will be used, regular communication, education and training.

    Unless you use AI to perform an isolated task, you’ll need to coordinate work and communication between AI and humans.

    “We have a very big group of people that dedicate themselves to legislative and privacy concerns so that the ‘big brother problem’ doesn’t become a reality. If a virtual assistant is spying on you, no one is going to use it. The symbiosis of man and machine is where we are going.”

    Rowan Trollope, CEO of Five9

    Help customers embrace AI

    Impersonal service and robot rebellion scare meat-bags

    AI not being able to personalize service scares people the most, among other things.
    Source: “What customers really think about AI: a global study”, PEGA


    “…it just takes one tiny miss for the customer to realize they’re engaging instead with a machine, and if they feel duped, you’ll never get them back for self-service, and the fallout could be much worse.”

    “AI in the Contact Center: Strategies to Optimize the Mix of Automation and Assisted Service” by Jon Arnold for Cisco

    Gain customer trust by:

    • educating them about what AI is and how it works,
    • showing them, how AI is used with their data, and that it is safe,
    • educating them that AI can learn from data on thousands of customers and offer something of value and personalized to each of them,
    • showing them, whether a human or a bot is talking to them,
    • guaranteeing that their data will be accessed only by need to know basis.

    “Consider that nearly half don’t understand that AI solutions enable machines to learn new things, and even fewer don’t know it can solve problems or understand speech.”

    Work on security concerns

    When working with and keeping huge amounts of data, it becomes vital to secure them.

    An equally important moment is to educate your customers that their data is secure and it won’t be used against them, because you use excellent data governance practices. “Don’t be evil.”

    Get the right AI talent

    Though big companies can afford to hire the rare data science and AI specialists with high-salaries, others might need to train the specialists inhouse – which isn’t that bad of an option. Thankfully, there are plenty of books, online courses and tutorials available to dive into the topic – it isn’t too hard to make the simplest of apps with the available APIs.

    It is also possible to consult with top AI experts on individual projects or an AI strategy.

    Mike Rollings, a research vice president of Gartner, says that to have adequate data by building a data-driven culture, chief data officers might need to work with:

    • talent sourcing,
    • skills development and training,
    • organizational structure,
    • analytical methodologies,
    • analytical tools,
    • data acquisition and monetization,
    • algorithm acquisition/creation,
    • analytical modeling,
    • analytical model training and maintenance,
    • process adaptation.

    They may also need to build a team of data scientists, engineers, statisticians and domain experts, who can apply AI to the business. Developing skills of current employees should also help with building and maintaining AI in the company.

    Read more:

    “Powering the service economy with RPA & AI”, BCG

    9. What to do after implementing AI

    Check AI’s health by checking the data sources you use and whether you have the skills to train the AI before and after deployment

    Clarify how AI will work to employees and provide necessary training, so they’d use the updated software.

    Measure, how well AI works for customers by timing interactions and recording calls – possibly use sentiment and tone analysis here as well.

    10. What Will Contact Center with AI Look like?

    The consultancy McKinsey & Company estimates that AI can automate as much as 45 percent or more of any particular job, allowing workers to focus on higher level mission-critical activities that cannot be as easily accomplished with technology.

    If the technologies that process and “understand” natural language were to reach the median level of human performance, an additional 13 percent of work activities in the US economy could be automated.

    According to the BCG Henderson Institute, despite the sensational headlines in the papers, AI isn’t going to cut a lot of jobs in the next 5 years. On the contrary, AI will take on unpleasant, simple, repetitive tasks people perform.

    Also, AI will work together with employees, granting them the computational power and insight they don’t have. To use this, employees will need to train how to work with AI in the workplace.

    There’s no doubt that AI will change businesses – the most reasonable way is to see through the hype and do the necessary work to use it in your company.

    Of course, one of the well known examples in retail is Amazon’s cashierless shop:

    Microsoft is going to develop similar technology for retailers as well.

    Perhaps, with enough IoT and AI in IoT, people won’t need to contact customer support?

    AI can reward customers for positive behavior, such as:

    • forgiving a mistake an AI agent has made
    • being polite with the agent
    • providing extra info that helps solve the case.

    Read an article by Dennis R. Mortensen on the topic, it’s quite intriguing.


    While starting to use AI may seem like a daunting task, it is only an illusion due to lack of information.. By learning more about AI, how and where you can deploy it to solve a business case in your company, you can begin your gradual adoption of this technology set to be ready to meet the current and future customer experience expectations.

    Hopefully, this guide has inspired you to use the practical approach to adopting AI and has shown you, how to make the process smoother and how to reach reasonable results.

    12. Read more

    Dive into these excellent resources to learn the nuances of developing and using AI in your company:

    1. “Reshaping business with artificial intelligence”, MIT Sloan Management Review & BSG, by Sam Ransbotham, David Kiron, Philipp Gerbert, and Martin Reeves.
    2. Putting AI to Work”, BSG, by Philipp Gerbert, Martin Hecker, Sebastian Steinhäuser, and Patrick Ruwolt.
    3. “Competing in the Age of AI”, BSG, by Philipp Gerbert, Jan Justus, and Martin Hecker.
    4. “Powering the service economy with RPA & AI”, BSG, by Philipp Gerbert, Michael Grebe, Martin Hecker, Olaf Rehse, Fabrice Roghé, Sabine Döschl, and Sebastian Steinhäuser.
    5. “The road to enterprise AI”, Gartner, by Whit Andrews.
    6. “AI in contact center and chatbot strategies”, by Jon Arnold, for Cisco.
    7. “What AI can and can’t (yet) do for your business” by McKinsey.
    8. “Four fundamentals of workplace automation”, McKinsey, by Michael Chui, James Manyika, and Mehdi Miremadi.
    9. “Notes from the AI frontier: Applications and value of deep learning” by McKinsey Global Institute – a more detailed PDF is here.
    10. “How to create a successful AI strategy” by Amit Ashwini.
    11. “What customers really think about AI: a global study”, PEGA.
    12. The section about AI in the enterprise at TechTarget.
    13. “How to Get Employees to Stop Worrying and Love AI”, by Brad Power, HBR.
    14. “2017 – 2018 Speech Analytics Product and Market Report” ($), by DMG Consulting.
    15. Open datasets for machine and deep learning.

    If you:

    • want to suggest an improvement,
    • got questions about AI in contact centers this guide hasn’t clearly answered,
    • want to share an exceptional resource about AI, an AI product or an AI API,

    comment below or write to me and I’ll update the guide.

    Thank you for reading.

    Start working through the steps to AI adoption to eventually use break-through AI products in your contact center.

    Share this guide with your coworkers to start building a team of AI enthusiasts to help you achieve this goal.


Leave a Reply

Your email address will not be published.


Got any questions?


Latest Entries

    Load more