eva 4.0 update: brand new features

Today we are happy to finally share the good news: the release of eva 4.0! our conversational AI platform  We have spent the last few months busy working on how to provide our clients with a better product and what new features and improvements would bring a better experience to your customers and, as a result, make a difference in your business. Several improvements and functionalities that will boost your project to an utterly new level, with a robust and scalable architecture. At the business level, it helps reduce time to market, as you’ll be able to quickly perform the deployment process and also speed up updating to new versions. 

 So, let’s take a look at what’s new? 

 

New design in our Conversational AI solutions

We know, we’ve been spoiling this one for a while… we were just too excited to keep it to ourselves! But now it’s official, eva 4.0 our conversational AI platform comes with a clean, fresh new look and some usability improvements that will make navigating the platform easier and more intuitive, always keeping in sight its well-known user-friendly quality. 

Organizations and Environments

 This resource will make the managing of Organizations and Environments a lot easier. It’s a perfect solution for those who have multiple virtual agents and can now navigate through them in the same space using the same login.   It also offers more flexibility to create different Environments (such as dev/test/prod) within these Organizations, according to the project strategies. Besides that, it’s possible to set different user access levels and grant permissions for each environment and the virtual agents therein. In other words: one same user can be an editor in environments A and B and a viewer in another environment C, for example. 

Agent Templates 

These are pre-built virtual agents that are ready-to-use, so you don’t have to start from zero (it may take up to 2 months of research, writing, building flows, testing). The Agent Templates are a collection of most common use cases by industry, starting from Banking to Foundation (basic flows), Healthcare, Help Desk, and Telecom.   It's a great way to better understand how eva works hands-on and inspire you build flows with the best practices in the market. Learn more about Agent Templates. 

Profiles and roles

To ensure a proper project management, it's important to have clear roles. Hence, we have updated the profiles and roles definitions in this new version to better respond to our clients’ needs. From two types in the previous version, we have now five different types: owner, admin, supervisor, editor, and viewer.    The idea is to allow a better understanding of the roles of each user in each project and, thus, define their access levels and permissions across all eva resources.  

Search within the Dialog Manager repositories

 When the project starts to grow and escalate, it’s just natural to have an extensive list of items on each repository in the Dialog Manager. To help users navigate through them easily, eva has incorporated a search function that finds specific cells (intent, entity, answer, service), flows, and AL documents or questions by typing their name on the search bar.  And we are not planning on slowing down, there is so much more on the way. Stay tuned for the great new features eva is bringing you in the weeks and months to come. [video width="2880" height="1640" mp4="https://eva.bot/wp-content/uploads/2022/07/13-search.mp4"][/video]  

When a Virtual Assistant goes public, companies face multiple questions that ultimately have to do with quality. How do I measure the quality of my conversational solution?

One way to measure the quality of our Virtual Assistant training is to apply an assertiveness measurement test.

Although the meaning of this last term expresses a social skill, it is currently used within the community to describe the ability of virtual assistants to give a correct or appropriate response to a specific question from a user who has expressed himself in a way that was not directly trained in the chatbot or virtual assistant. There are several ways to correctly measure assertiveness, but they can be grouped into three main ways to measure that increase in complexity and cost.

1. Indirect rate of assertiveness:

When we talk about a fallback, we are talking about a response where the assistant was not trained and responded with a message like "I didn't understand". In this way, you can create the easiest indicator of assertiveness, which would be to take the total number of fallbacks and divide it by the number of interactions that came into the bot during a period. This would give a fallback rate, and its complement would be the assertiveness, so we are talking about an indirect assertiveness rate. It serves to roughly know how much volume of questions are coming in that the bot has not been trained for, answering that it doesn't understand.

2. Strict assertiveness rate:

At the other extreme, the most complex way to measure assertiveness requires the common agreement of two or more parties that select a representative sample of inputs or real examples of users with which the system will be measured and then manually annotate each of the inputs with their outputs, i.e., the response that the system actually gave, and identify whether the sentence belongs to the bot's knowledge domain and whether the classification or response it delivered was adequate or not. Once the group of annotators has made the relevant evaluation of the same set of data, the degree of agreement among them is evaluated, because it is possible that some of them may have considered that everything was relevant and adequate in a random way. A simple statistical test allows to solve that, creating an annotated collection of great value for further training improvement. The work is cumbersome and time-consuming and even requires some training for the annotators. This way of measuring the Strict Assertiveness Rate is recommended only in cases where the indicator is linked to some obligation that requires formal demonstration.

3. Semi-Automated Assertiveness Rate:

An intermediate approach is the Semi-Automated Assertiveness Rate calculation procedure, which saves time and is often an ideal formula in agile contexts where the quality of our Virtual Assistant needs to be measured and updated by demonstrating its value. Depending on the type of conversational solution, the calculation will be made by first identifying all the training, linking it with the answers that will be measured. With this input, a table is generated where the actual sentences and the response that "should" have been received. This task is usually abbreviated by simply using the intent that should have classified that sentence. Because in practice manual effort is usually required in this part, the "semi" part of the indicator's name comes up. In some cases, it is possible to automate the entire flow from start to finish, but there are often conditions that make this task difficult. Then, a second external bot will "send" the sentences to the virtual assistant. The wizard will respond with its answer and that answer will be saved, giving rise to a data collection containing each of the actual user inputs, the classification that should have been delivered and the classification that was delivered. Finally, a matrix is created with the frequency of correct and incorrect classifications, thus creating the assertiveness rate indicator par excellence, which allows us to identify with a good level of detail and relatively quickly which are the knowledge domains that the bot does not handle and in which the training fails more in a familiar indicator expressed as a percentage. The first insight we have seen generated in these measurement experiences is the need to merge some answers together, to avoid confusing the dialog engine that runs the wizard. There are an infinite number of ways to combine these measurements and the three levels are rather didactic to describe their complexity. Usually, more steps are added to the measurement as each virtual assistant's own requirements emerge. Having a proper measurement of the assertiveness of our bot will ensure its quality with the support of an indicator that impacts the user experience and the final evaluation of the virtual assistant. With the measurement comes a subsequent re-training process that must be carried out carefully to avoid diminishing the generalization capacity of the model on new cases for which it was not trained. Another interesting read: A virtual assistant said: I’m sorry, I didn’t understand correctly, I’m still learning, can you write it another way?  

Natural language understanding (NLU) refers to the use of artificial Intelligence (AI) in the form of machine learning (ML) to allow computers to understand the meaning of human language, commonly known in the context of computer science as natural language. Albeit complex and ambiguous, natural language shows implicit rules and patterns that can be tackled through complex statistical models. By training machines on vast repositories of curated linguistic data, we can use machines to associate any given linguistic inputs or utterances with one of a set of predefined meaning categories, usually known as communicative intents. Understanding the meaning of an utterance, in this sense, is understanding its function in the context of communication, what the language does.

Conversational AI

Conversational AI leverages accurate NLU capabilities with a dialog management component to build conversational digital interfaces, that is, digital interfaces that use natural language as the key form of human-machine interaction.

Why is this so remarkable?

Well, it is said among designers that the ideal interface is one that is not noticeable at all. An interface that blends with the task at hand, and, without any undue friction, allows you to achieve your goals. Language, if you think about it, is one of the best exemplars we have of such a concept. It allows us to connect with others and shape ideas, as well as our world, while remaining in the background.

Conversation as technology-empowered language

The exponential development of NLU and Conversational AI technologies has anything but challenged the key role of language in our daily interactions. Rather, it has empowered our words, opening a wide range of new ways and contexts for it to transform our human-to-human communications, and, most notably, the way we communicate with the machines that characterize our world today. Conversation has been recognized as the key format of the digital world. As places made of language, digital systems have been designed to talk with users in a language that’s in tune with who they are. Some researchers have even gone as far as to affirm that the language of the digital age represents a sort of come back of the spoken word, a second orality, while others have described the steep increase of communication via instant messaging apps, texting, as an emerging kind of fingered conversation. Conversational AI allows organizations to use language, the most universal human interface, as the building block of effective and efficient customer journeys. While statistics and ML are used to leverage the subtleties and complexities inherent to language understanding, empathy and creativity are crucial to better understand your users’ needs and guide them along carefully designed conversational paths, streamlining every conversation between your organization and your customers.

Want to learn more about Conversational AI?

Don’t miss Santiago Santa María’s Masterclass on Conversational AI, where you’ll learn how companies across different industries have transformed their operations thanks to automated and AI-powered conversations.  

For the very first time, ARCOmadrid has a Virtual Assistant to enhance the visitor’s experience. In it’s 2022 version, ARCOmadrid, the Spain’s international contemporary art fair, has surprised its public with a revolutionary demonstration that is not part of the exhibition itself. We are talking about ARC, a friendly Virtual Assistant whose job is to complete the visitor’s experience. By using conversational AI, ARC can give you information about the exhibitor galleries, the exhibition sections and even about the architecture of the venue. Also ARC can interact with any visitor, it talks directly to them, anticipates the user’s needs to suggest them what to do, and it even changes from male to female randomly. Two major companies have come together bringing the art to the digital era: NTT Data, 6th best IT company in the world, and LLYC, lead communications consultancy company in Spain. They have worked tirelessly for months to bring ARC into the fair, using NTT Data’s own enterprise conversational AI platform, eva. eva is an enterprise conversational AI platform that helps you create Virtual Agents from scratch. Powered by it’s own NLP (natural language processor) eva has been able to take ARC beyond the robot, and bring a smart digital person into the exhibition. This powerful technology opens up a world of new possibilities that can be applied in any physical space in any industry. A good example of how we merge the physical and digital world creating phygital spaces. ARC, that has been placed in a totem at the entry of the fair, is the proof that technology can improve any sort of experience in any scenario. This experience proves that this technology doesn’t belong exclusively to the tech industry, and it can be used in any environment. This is the first step of many. ARCOmadrid has opened the door to endless possibilities of proving that communication can be taken into the virtual world, and that information may be provided in the physical world in a more graphic and dynamic way. NTT Data is proud to be a pioneer, bringing this technology into the art world, and being the first company in placing a Virtual Assistant in ARCOmadrid’s history.  

The alliance of NTT Data and Voximplant announce eva Voice Cloud, a powerful technological solution that offers digital transformation in call centers to companies, with a 100% cloud solution that revolutionizes user experience. NTT DATA, a leading company in transformation, technology and consulting operations in Europe, the United States and LATAM, strengthens its alliance with Voximplant, a leader in cloud call center solutions. The partnership to revolutionize call centers with Conversational AI will be through eva, the platform to create and manage virtual agents created by NTT Data experts which now extends its capabilities in the call center with Voximplant. Together they will offer the best cloud solution to transform the customer experience in call centers. Companies understand they must ensure a high level of user experience, and need to react robustly to new market conditions. User satisfaction when interacting with call centers is very low. Most companies fail to offer a call center service that meets the customer's expectation. Strategic alliances such as  NTT DATA and Voximplant, aim to shed light on this need with an agile and efficient solution. "The most important thing for us is agility, being able to deploy call center solutions in short times where we automate phone calls," says Santiago Santa María Director of Conversational AI at NTT DATA. We created eva Voice Cloud powered by Voximplant, a robust and  complete business solution, which uses artificial intelligence to improve the user experience in voice conversations. "Collaboration is born from the need to drive improvement in care; the pandemic has forced companies to make courageous decisions. Today, migrating to the cloud is a necessity that accelerates the transformation of customer service," says Santiago. No doubt remote work and social distancing increased call center calls.  eva by NTT Data provides robustness and quality in the cloud, thus allowing a faster dialogue with customers. This alliance is ambitious and promising, as both companies have a wide commercial expectation in the use of AI in call centers. Responding to the user immediately and solving their need the first time is a priority and a fundamental requirement to compete in the market of today and the future.

Creating virtual agents with eva is now faster and easier with the new templates we have designed for you. What can take up to two months of development between research, writing, flow creation, testing and more, can now be reduced to less than a week with Agent Template's.

Agent Template is a collection of flows provided by eva that can be used to establish a base for building conversations.

Currently available:

• Banking Agent Template, with 37 ready-made flows and 5 use cases for financial services. Available in english, spanish and portuguese.
• Foundation Agent Template, with 13 flows common to many industries and sectors, such as NPS, Welcome, and Talk to an Agent. Available in English, Spanish and Portuguese.
• Healthcare Agent Template is a collection of 18 flows focused on healthcare service that can be used to establish a base for building conversations. Available in English, Spanish and Portuguese.
• Ticketing Agent Template is a collection of 19 flows focused on Ticketing service that can be used to establish a base for building conversations. Available in Spanish and Portuguese.
• Telecom Agent Template is focused for Telecom virtual assistant, featuring 25 flows. Available in Spanish and Portuguese.

They were designed based on market best practices, with the goal of optimizing both your team and the process of building a virtual agent. We worked together with people from NTT DATA and Digital Experience. 

In agent templates, you will save time on your project because it offers:

• Ready editable flows. In other words, your team won't have to worry about building the bot flows. You only need to edit the steps and adjust the responses to your business needs.
• A bot ready for each industry, with specific use cases.
• Text and walkthrough suggestions, material developed with UX best practices and various studies (such as benchmarks, interviews and user testing).

The consumer will only have to customize the information and webhooks according to their business case.    We have created three versions: Spanish, Portuguese and English, and a manual to help your team.    Agent Templates can speed up many tasks and allow teams to act more strategically. New agent templates for different industries and use cases will be added soon You can download the agent template on Github You may be interested in reading this article: 4 reasons for banks to bet on AI Conversational

I ask the Virtual Assistant: Are you really learning? Is someone helping you learn? Do your creators know how to teach you? Do they know what to teach you? I don't think you have learned yet that we, the users, are desperate to save time, avoid long physical lines or tedious minutes of waiting while a call center executive attends to us. When we contact you, we are really impatient.

Let me tell you a story of an experience with a virtual assistant.

Yesterday at 10 o'clock at night I had no TV service, just when there were no branches open, no call center available. And the only thing I found was a floating virtual assistant button, and I couldn't get through the maze of your training to avoid the dreaded "I'm sorry, I didn't understand correctly" sentence. And without being able to help me, you also give me no option to talk to a human who can understand me. You think my satisfaction is increased by just having a floating chat button, a cute name that has the letters "AI" and the best designed avatar.  I think I and many users expect your "AI" (Artificial Intelligence) to be really smart enough to be able to detect that we need to solve problems quickly. I must admit that when you understand me, I really solve my problems in seconds and not in minutes or hours! That's why I want to help you to solve more, and that your "I'm sorry, I didn't understand correctly" is not the sentence that makes you sadder for not being able to help a user with needs. I imagine you sad and overwhelmed, even a little frustrated for not being able to help more. And at the same time, eager to be helped. Help you to be a key player in increasing users' digital NPS, to be central to your company's cost-saving strategy, to make more users prefer to tell you what's wrong with them, and to increase your success rate.

If you can, tell your creators what you already know you should do:

1. Always convey that you're a bot and not a human. Don't try to look like a human, because the differences are still noticeable.
2. Teach me the best way to write you, so I can increase your assertiveness rate.
3. Let me go back so I can ask you about something else. Don't expect me to just end up where you imagine, because I will most likely want to resolve other things.
4. Don't try to be omnipresent. Surely there are places on digital sites where you simply do not help. It is better not to be in those cases, because I get frustrated and lose the desire to continue the journey.
5. If you are not yet trained to solve some issues, give me the option to talk to a human. Relax, you've already solved many transactions and that's fine. But sometimes you have to make room for someone else.
6. And if you can refer me to a human, don't be mean and share all the information. Don't make me waste time telling the whole story again.
7. Finally, we both know you have all my data, you know what I wrote, at what time, you know if I have already entered other digital channels, even if I have gone to the branch or called the call center. I imagine you might know if I am a good and profitable customer, or my ARPU is higher or lower than average. Ask to look at metrics and statistics so you really learn.

How do I know all this? At NTT DATA we have a methodology that allows us to achieve virtual assistant assertiveness rates of 85% or higher. Our OKR methodology based on metrics and digital analytics, allows us to have a set of Performance, Experience and Business indicators to achieve ROI and increase profitable customer experience.

Our goals are ambitious

We have even set out to achieve in the next 6 months of the launch of a virtual assistant:

• Obtain 85% NPS
• Decrease Churn rate to 35%.
• Decrease the rate of forced executive handovers to 30%.
• Increase Bot accuracy to 85%.
• Decrease false positive rate to 30%.
• Increase case resolution by 85%.
• Increase positive feedback percentage by 85%.

Let me help you not be embarrassed to say "I'm sorry, I didn't get it right, I'm still learning, can you write it differently?" You may be interested in: 5 steps to writing accurate virtual assistant scripts

The digital transformation of traditional businesses has been a priority in recent years and has been challenged and even more relevant in recent times, because of the hyper-digitization necessary to sustain business due to the pandemic we are still experiencing. Digital transformation leaders have been invaded by an unprecedented demand for digitalization initiatives. This has involved different strategies for prioritizing and satisfying needs in a short period of time. It has undoubtedly put in value the structures and models of digital transformation adopted by each company.

All initiatives share at least one of the following drivers of digital transformation:

• Increased customer experience
• Internal efficiency
• Increased digital conversion and digital onboarding of new customers. 

Therefore, we have seen how many companies have been convinced of the relevance of digital transformation for the survival of their business. For traditional companies, we have seen a dilemma for the survival of their business. Taking care of the market share of established companies among their traditional competitors and protecting their market from new entrants (mostly born-digital) has put digitalization initiatives on top priority.

We often see the user-centric digital product design model increasingly adopted, at least in intent.

 The user-centered digital product design model is nothing more than putting users' needs for a digital product first. While simple to say, executing it is very complex and requires a lot of discipline. To do this, Customer/User Experience experts use multiple tools to obtain insigth from users to define which are the main needs to prioritize in each of the strategic lines of the company. The change of mindset is very important, and I give a brief example. Lately we have seen an explosive increase in Digital Wallet solutions, mobile payments, QR payments. It seems simple to imagine that a Digital Wallets product requires a payment method enrollment functionality (like a credit card) and as that functionality those who think and design the product can define hundreds of other functionalities.

Now, the above reasoning is product-based and not user-centric.

Expanding the mindset and applying user-centered design we can imagine that the user wants is:

• Looking for something you need to buy
• The user will need to pay as quickly as possible.
• They will want to pay as securely and reliably as possible.

With this mentality it is possible to detect needs that is not simply to enroll a means of payment. And thus, focus on what the user really needs to satisfy to build memorable digital products. We have seen how the production of new features of digital products has increased exponentially. And at the same time, we have seen how many of these features are not well received by users because they have been designed with a mindset focused on the product and not on the user. The increased budget invested in many new digital functionalities that do not fulfill the promise of improving any of the three digital drivers indicated at the beginning (Increased Experience, Efficiency, or Increased Conversion) is very high and the return on investment is often not achieved. In my opinion, not achieving the expected ROI of many of the new digital functionalities is due to not thinking, designing, and prioritizing the product focused on the user. But even more relevant is the overcoming of digital transformation leaders by the beastly functional competition between companies. That is, the prioritization of digital functionalities is not by product-centric prioritization, nor user-centric prioritization, but the prioritization of new digital functionalities is simply to match the new digital functionality that the competition has launched in the market. The competition between short-term objectives of cost reduction and market share growth and the incorporation of digital functionality just to match the competition overshadows the real needs of digital users and customers. Digital transformation leaders must advocate for positioning or repositioning the user-centric product design mindset in their teams. To provide an optimal experience in digital transformation, we invite you to learn about virtual assistants. Great experiences to transform your business