Tibor Vass (00:00):
We are on time, and thank you very much to everyone who joined us today in this webinar series. We’d like to start our discussion about a recent topic: how generative AI revolutionizes case management. We prepared a demonstration and a couple of slides to show how Eccentex is approaching this exciting opportunity and what we are able to do.

Tibor Vass (00:30):
My name is Tibor Vass, CMO of Eccentex. I will give this presentation together with my colleague Maksim Gill, who is the Vice President of Product. He has prepared a demonstration where we will show how this technology can be applied in day-to-day business. Without further ado, I would like to start my presentation with this slide, which is important for defining what Eccentex is today.

Tibor Vass (01:04):
Most of you probably already know at least a little about Eccentex. We are a low-code platform, and we own innovative technology that helps business users create their own applications by leveraging specific builders and designers, which make application development very easy. In recent months, we have aimed to move beyond low-code technology to the next generation of business application development, which we call no-code.

Tibor Vass (01:42):
We took a big step forward by employing new AI technologies that enable us, in certain cases, to create business applications without writing any code. We have already deployed several Eccentex AI applications that help business users adopt new functionality with a single click, and we will see Max demonstrate some of those new features.

Tibor Vass (02:17):
All the applications we are building will be equipped with new embedded AI technologies. They help customers and business users search, analyze, or even reply to certain cases, and can even invent a new solution with a single click.

Tibor Vass (02:35):
This low-code technology is extremely important, especially in the near future. Forrester CEO George F. Colony recently made a bold statement at an innovation and technology conference: the most popular programming language of the future will be English. It is a very bold, but most probably valid, statement. Eccentex is adapting its development and technology roadmap accordingly, and our aim is high.

Tibor Vass (03:12):
We’d like to help business users apply and deploy technology and business applications using no-code about 60% of the time, and use additional low-code platform technology to augment or tailor the solution about 35% of the time,

Tibor Vass (03:33):
with deep code needed only about 5% of the time for further customization or specific application interfaces. That’s our aim. We know it is very challenging and ambitious, but this is the roadmap we’d like to follow over the next couple of years. Very importantly, we came up with a new offer called HyperAutomation Cloud, based on a redefined concept of hyperautomation, which Gartner originally defined in 2020 and called the number-one trend in strategic

Tibor Vass (04:15):
technology at that time, and it’s still valid, and our definition is close to, but a little bit more than Gartner defined at that time, because we do believe that HyperAutomation is definitely not just RPA and adding an AI capability to the

Tibor Vass (04:33):
robotic process automation features. It is an orchestrated use of multiple technologies, tools, and platforms, such as AI, machine learning, event-driven software architecture, business process management, operational CRM, knowledge bases, and automation. Together, these define a new way for HyperAutomation to deliver. We are serious about executing this, and our first step a couple of months ago was successfully trademarking HyperAutomation as a Service, so we

Tibor Vass (05:09):
own this new business category. We would like to make business users’ lives even easier, not just by allowing them to use our platform to design, create, and deploy new business applications with embedded AI features, but by operating those applications on their behalf. Our aim is to deliver customer experiences on behalf of our customers to their end users by using our technology platform and adding a service on top that includes

Tibor Vass (05:49):
operations and maintenance, as well as application development services. That’s why we created and deployed HyperAutomation as a Service, which is available to companies that would like us to handle IT operations.

Tibor Vass (06:07):
How does it all work together? Those who follow us may already have noticed that we launched two new platforms and domains, eccentex.ai and eccentex.io, on top of eccentex.com, our traditional platform. We did this because we are segmenting our services into three strategic categories. Most importantly, eccentex.com is responsible for all services and publications and for the platform that operates as the core of the HyperAutomation

Tibor Vass (06:43):
concept. The professional services part of our business handles integration, deployment, and configuration services, while also exposing open API connectivity through the eccentex.io domain. Our eccentex.ai domain collects and offers the cloud operations and AI operations services included in our portfolio. We would like to deliver all these services together in a highly orchestrated and integrated way.

Tibor Vass (07:19):
HyperAutomation Cloud as a concept contains the software, related services, and all the AI features attached to them.

Tibor Vass (07:32):
I know this may be too busy, but I would like to highlight the platform and the capabilities we have developed so far. As you can see in the middle, the Eccentex AppBase platform is still the core of the offer. It enables customers to create, deploy, and operate their business applications, and it enables us to store business applications we have already developed and offer them as core applications, such as email automation, dynamic case management, knowledge management, and help desk. These applications are horizontal by nature and enable business users to deploy them even in vertical use

Tibor Vass (08:14):
cases. On the left is the innovative platform we created through eccentex.ai, which is a federation of different third-party and in-house AI services. We do not want to compete with or go

Tibor Vass (08:34):
against best-of-breed AI services available on the market; we’d like to use them, integrate and orchestrate them together, and expose them as one service to our customers. That’s why we created a personification of our AI called Eccentex kAI. It is an interface through which customers can leverage different third-party AI services together with our AI capabilities in an orchestrated way, without having to consider

Tibor Vass (09:09):
the different background integrations needed. We are doing this because the AI services available on the market have different core capabilities and strengths. We select the best-of-breed AI capabilities, put them together, and expose them as one service to our customers. On the right side is eccentex.io, the open API gateway responsible for integrating our platform with every component of the CX ecosystem, such as contact centers, third-party CRM and workflow systems, databases, knowledge bases, and whatever else is

Tibor Vass (09:48):
needed to provide the customer experience. We do this because we believe that customer experiences exist at the ecosystem level. Nobody in the market can deliver these experiences alone.

Tibor Vass (10:01):
We need to work together and orchestrate these capabilities. That’s why we opened our platform and enabled third-party integration. We strongly believe that the Customer 360 desktop, which sits in the middle of the experience and gives the agent a consolidated view of the customer, is very important. It requires significant integration because the data components needed to serve customers during their journey

Tibor Vass (10:31):
reside in different systems and need to be pulled together in real time, integrated, and displayed intuitively on the desktop. Without further ado, I would like to go to my last slide, introducing eccentex.ai. This new platform component enables our customers to engage, automate, and empower their customers and agents with our AI capabilities.

Tibor Vass (11:02):
In the next couple of minutes, we will see a demonstration from Max of how we deployed many of these capabilities. Because time is limited, we decided to present almost half of them in this solution demonstration. You will see some great examples of how we apply our capabilities to enhance a real-life use case and take the customer and agent experience to the next level using AI.

Tibor Vass (11:40):
I would like to hand over the screen to Max for the live demo. Max, I have stopped sharing my screen, and you can take over.

Maksim Gill (11:51):
Can you give me just a few minutes? My internet just went out all of a sudden. I need a few minutes to restart it. Give me a few minutes.

Maksim Gill (12:01):
Apologies.

Tibor Vass (12:02):
No worries. I’ll speak about our AI Services while Max restores his internet connection. The AI Services we developed are based on real customer needs and real-life use cases. We didn’t develop them using a central, inside-out approach; we took external AI use cases that customers need to deploy based on real-life scenarios.

Tibor Vass (12:38):
Many customers who contacted us struggled to maintain a large volume of emails that all needed to be answered in a hyper-personalized manner. That’s why we developed solutions that decrease the time needed for a sophisticated and adequate answer from eight to ten minutes to about 40–45 seconds.

Tibor Vass (13:07):
We also use the same AI capabilities to identify and mask PII—personally identifiable information—from an email. That information should not be displayed to an agent, but it still needs to be captured and processed securely. We allow agents to automatically classify every incoming interaction without following a script or engaging further with the customer. We can determine the priority based on the free

Tibor Vass (13:42):
text in a case or email, assign the priority, dynamically search various databases and knowledge bases, and immediately assign the best possible answer or

Tibor Vass (14:02):
response to an incoming case automatically. We are also concerned with helping customers assign different business rules—which can sometimes be complex—to certain cases, ensuring that nothing is forgotten. Everything will be performed on time, and every customer will receive their answer or service when promised.

Tibor Vass (14:36):
We also figured out how to create cases automatically from email. This makes agents’ and customers’ lives easier because agents do not need to translate the content of the email into a case. We can extract the most important parameters automatically and create a case without human touch, while still allowing an agent to oversee and modify it as needed.

Tibor Vass (15:07):
We can also perform sentiment analysis on emails, other written media, and transcripts that go through our concierge bot. I won’t introduce more details because Max will do the demo. Max, are you ready? Yes. No worries; technology is always doing this kind of thing. I have stopped sharing, and you can take over the screen.

Maksim Gill (15:35):
Can everyone see my screen? Yes? The right one here with the email? Yes. Okay. I want to discuss the various concepts we have around AI.

Maksim Gill (16:00):
Let me refresh here for a second. The idea behind our case management system is that everything is created as a case. This is an example of how a case management system would look inside a contact center—Genesys in this example—but it can work with almost any contact center or stand completely alone. I like to start with email cases because they provide a good introduction to natural language processing, AI, and other capabilities.

Maksim Gill (16:32):
Let me take you through this analysis. A customer sent an email to a customer service help desk, and the system converted the email into a case. This is important because the case lives throughout the entire customer journey lifecycle. As the conversation goes back and forth between the customer and agents, a chatbot, or whatever other channels the customer

Maksim Gill (17:03):
uses to interact while resolving their problems, everything is recorded in one case. This provides far more AI and machine-learning capabilities than a transactional conversation. Everything is connected, and you can determine later how it was resolved. Here we have an email in which the customer asks several questions. They are trying to switch from a Samsung phone to an iPhone, have several questions, sent other information, and want to order the phone. It is a fairly complex email.

Maksim Gill (17:41):
The first thing we do with AI is choose among several engines to use the best one available. Our primary, go-to AI engine—the one most of our work is built with—is Azure AI. One reason is that Eccentex Cloud is primarily powered by Microsoft Azure. For security and enterprise use cases, Azure remains one of the best platforms for our purposes and provides us with many AI capabilities.

Maksim Gill (18:20):
And so we’re analyzing a bunch of stuff here so we’re extracting product names. This is important for routing, we can then route this information to the right product team.

Maksim Gill (18:31):
We can identify the language and urgency. This is all done in real time. The urgency is high because they want the phone by July 17, and that date has already passed. We can translate into more than 100 languages. We do not translate only sentence by sentence like a traditional translator. We use a neural network through an LLM, which can also translate the meaning.

Maksim Gill (19:06):
It handles not just each sentence, but deeper meaning and nuances in the language. We can demonstrate that a little later. We can summarize either the customer’s original question or the entire conversation. As the conversation expands, more topics can be covered. You may originally ask how to perform technical tasks, then end up asking billing questions

Maksim Gill (19:38):
or about a previous case. All of this can be summarized so the agent can quickly see what is going on. Sentiment is also important: angry, unhappy, or positive. We indicate not only whether the sentiment is positive, negative, or neutral, but also percentage scores—in this case, completely neutral—which can later be used to fine-tune

Maksim Gill (20:02):
AI. Then there is topic and tag detection. The system can automatically generate a list. You can predefine the tags yourself, or we can run our AI engine through hundreds of thousands of previous cases and provide a list of 200–300 topics that are relevant and common across those types of cases and

Maksim Gill (20:33):
emails. We can provide these categorizations, helping you understand the topics without having to predefine them yourself. We provide many of these utilities out of the box to simplify your work. You do not have to spend a lot of professional services time figuring this out and performing the analysis; we do much of it for you.

Maksim Gill (21:00):
Then come what I call the color-coded items, which are specific to your use case. This is called entity extraction, and it uses AI through an LLM to identify items important to you. For this use case, it was important to identify people, contact information, addresses, organizations, events, and products. These are highlighted based on the use case. For a different use case, such as a car dealership

Maksim Gill (21:38):
or warranty service, I might want to extract VINs, car models, and other vehicle features. For a hospital or healthcare company, I might extract information such as symptoms and medications. This is all configurable within our product and uses a very powerful engine. If you are extracting an account number, you can specify a pattern. Otherwise, you can ask for a list of all the people,

Maksim Gill (22:12):
and it uses a large language model—essentially GPT-4—to understand who is being discussed and identify those people. Other useful AI features include PII detection and redaction: the ability to extract credit card numbers, passport numbers, and Social Security numbers.

Maksim Gill (22:40):
You can go further and extract things such as people’s ages or symptoms. In this use case, the person sent a credit card number. Before it reached the agent or user, we were able to redact that information.

Maksim Gill (23:02):
It can be completely redacted from all system records or simply hidden from agents. If an agent or supervisor wants to see it, they need to click a button and approve access, and the system records that they viewed private information. As the conversation expands back and forth, we continually add more topics and tags to the case.

Maksim Gill (23:37):
The final capability in this agent view is Smart Knowledge Base. For those who do not know, our product includes a built-in, full-featured knowledge base system as an add-on. It is very powerful for enterprise use cases. We provide smart, context-centric search for everything relevant to the case. It can be used like a simple Wikipedia, where a user searches for content or links,

Maksim Gill (24:13):
but the more relevant use case is highlighting or suggesting articles relevant to the case. We consider the topic and what the case is about. This information can come from case-specific details.

Maksim Gill (24:35):
For a fraud investigation, the system knows what the fraud is about and has its description. In this email use case, it understands the questions and begins suggesting the most semantically relevant articles. But it does not rely only on semantics. We also know the language and have information about the customer: who they are, their

Maksim Gill (25:03):
priority level, where they live, and their previous products. We will not suggest warranty centers in Florida if the customer lives in California, even if they did not mention their location in the case. We can contextually feed this information to the search engine so it provides highly relevant results. We also have Smart Replies, which can automatically generate a response to a customer’s question.

Maksim Gill (25:40):
We have a complete template and snippet system, as you would expect in an email help desk. The agent can write instructions describing what they want to tell the customer. In this situation, the agent wants to explain how the customer can accomplish what they requested.

Maksim Gill (26:00):
The system generates a response fairly quickly. It creates a draft that goes through all the customer’s questions and answers them. I can change the prompt—for example, to include a FedEx tracking number or ask for additional information—and it automatically generates the response. The agent can then make changes as needed. The system can also be configured so that an agent’s response to a customer’s request passes through its own AI engine.

Maksim Gill (26:42):
For example, if an agent starts swearing, becomes rude, or introduces another issue, the system can detect it before the response is sent. Because of neural-network translation capabilities, the agent can also respond in the language they prefer.

Maksim Gill (27:06):
The system can then generate a language-specific response. Our product includes many NLP and NLU—natural language processing and natural language understanding—capabilities for simple inquiries, chatbots, and similar uses. The product also includes its own CRM system.

Maksim Gill (27:32):
We have a full-featured, service-centric operational CRM capability. It is configurable, and our system integrates with other core systems. If you want to bring in customer data from SAP, Salesforce, IBM, or another core system, you can combine all that information on one screen.

Maksim Gill (28:00):
Since this discussion is primarily about artificial intelligence, I will focus on that. This is a use case for a retailer such as Costco. We can see all the customer’s previous orders and detect what they may be calling about. Before even picking up the call, the agent can see highlighted information that is probably important.

Maksim Gill (28:37):
So, we know that the person may be calling about an email that the customer sent a bit earlier or a late delivery that has happened. So, again, using artificial intelligence, we’re able to highlight things or surface information that is very relevant to the customer journey here predicting

Maksim Gill (29:01):
what they’re talking about ahead of time and trying to get the agents acquainted with the situation so that everything is done faster and more smoothly and that the customer feels more personalized if actually someone knows what the problem is. Another part of the AI right here is the ability to use all sorts of information we have about the customer and all the previous inquiries or cases, whatever happens to provide maybe offers or next best actions.

Maksim Gill (29:32):
And these are, you know, in this situation because I kind of made this in more into sales use case, we have here offers. So, we know what the person has purchased because we integrate with whatever core systems that the customer may have. We also know all the previous email inquiries we know the topics they talk about, you know, we know how the previous case resolved. So if a customer is filed a complaint against, you know, about internet speed in a travel that they recently had, then we can make sure that we don’t offer that type of service again.

Maksim Gill (30:12):
We can apply machine learning to all previous interactions with the customer, as well as demographic research. We know, for example, that this person is a Gold member, how many years they have been a member, and that their spending this year is less than a quarter of their typical annual spending. We can then provide more relevant offers. If they recently bought a TV,

Maksim Gill (30:44):
and they are the type of person who might buy an extended protection plan for it, or if we know they have a spouse who may want a second membership, we can present those personalized offers. We can do this because all the information is consolidated in one place and can be analyzed with machine learning.

Maksim Gill (31:15):
How was all this built? It looks tailored for a specific use case, and it is. Behind it is a powerful low-code platform for case management and operational CRM that allows all of this to be configured

Maksim Gill (31:35):
relatively quickly. In the setup, we can design the customer model. Customer information is modeled here, and each type of case you want to handle—such as a fraud investigation—can have its own data models, workflows, and other elements.

Maksim Gill (32:05):
You can model the entire database or customer without requiring all the information to live in our system. The data can be synchronized with our system, or we can call core systems to retrieve it as needed. Because it is all designed here, AI and machine-learning use cases become much easier. We can assume it is all available.

Maksim Gill (32:37):
AI results can be shown anywhere in the system. For example, the Eccentex AI widget you saw for offers can be dragged and dropped into place and configured. We have workflows and many rule builders that can use AI extraction methods. Here is a simple AI triage rule builder. If an email or case with specific information arrives, we can define conditions such as the language code being English,

Maksim Gill (33:20):
the sentiment being negative, or the partner or customer being in network. By designing these conditions, we can decide what we want the system to do.

Maksim Gill (33:31):
We can increase the case priority, send a special VIP acknowledgement rather than a standard one, and assign the case to a VIP team. Using sentiment, language codes, and other indicators or predictions about the case or customer, we can configure actions very easily.

Tibor Vass (34:01):
I would like to add an important point. This is a great example showing that AI analytics capabilities are integrated directly with the rule engine. Whatever AI identifies in the text or during the analytics phase can trigger an action in the rule engine. It is all integrated within the system.

Maksim Gill (34:23):
Yes, absolutely. The AI boom has provided many capabilities, but there is always a question: how do you actually use them in the real world?

Maksim Gill (34:38):
You can integrate ChatGPT and have it create knowledge base articles, but you could also go directly to ChatGPT, write what you want, and copy and paste it into your knowledge base. We provide value because all the information about the customer, cases, system context, and current rules is in one place. The content we generate, rules, UI, and machine

Maksim Gill (35:11):
learning are all based on the contextual data we have. This provides real use cases that matter to you rather than simply copying and pasting between systems. We can now transition to one of our newest and most exciting features: Solution Ventor, which uses a narrative to build a system like ours.

Maksim Gill (35:46):
As you saw, our system has many configurators, page builders, and workflow builders, but configuring them takes some work. We created the ability for a customer, user, or administrator to explain what their case management

Maksim Gill (36:04):
system should look like. You can enter up to 20 pages of text, copy and paste an RFP, or describe it however you want, then view the results. Here I say that I want to create a complaints case management system for a clothing retailer. It can generate several case types, though I asked it to generate only

Maksim Gill (36:32):
one: a product-quality-complaint case type. It generates the workflow, explains each element, and defines resolution codes, form fields, the data model, forms needed to initiate the case, and the people involved, such as the complainant, case manager, and product quality investigator. It creates operational CRM data models, forms, search screens, reports, and more.

Maksim Gill (37:03):
It also generates reports, such as a complaint summary listing complaints, related product categories, customers with the highest number of complaints, and their categories. You can create these reports based on the case types, along with other items such as email templates.

Maksim Gill (37:35):
These can be attached automatically to workflows. For example, a complaint-received letter template can be attached to a milestone so that the letter is sent when a case is created. This gets you far ahead. Except for a very simple business, it usually does not generate an application you can put directly into production, but it gets you close and provides a strong starting point, including the interfaces and

Maksim Gill (38:06):
data models you need to begin prototyping. You can experiment, explain your use case in more detail or differently, and change things so it fits your needs. I have hidden the button in this demo because people keep pressing it, but normally there is a Create Entities button that creates everything in our system. You can then use the various builders to adjust it as needed.

Maksim Gill (38:37):
We have many AI capabilities on our roadmap, including advanced document processing, better language understanding, text labeling, and other machine-learning features. These will be a core focus of our product going forward.

Maksim Gill (39:03):
Everything will be AI-enabled. We are eager to work with customers on specific use cases. We have done substantial work integrating these AI capabilities into our product while keeping it flexible. If you have a particular use case, feel free to reach out; we can configure the product to handle it.

Maksim Gill (39:34):
We have five minutes left, so I’m going to hand it over to Tibor.

Tibor Vass (39:38):
Thank you, Max. It is gorgeous. I am always learning, even though I have seen this presentation a couple of times. This is great because new AI features are added to our system almost weekly. Can you see my screen? This is how you can engage with us. If you have any questions, please email us, visit our website and submit them there, or call our sales team directly.

Tibor Vass (40:15):
You can follow us on LinkedIn and ask a question through LinkedIn or directly from Max or me. Please visit our website and consider how the use cases Max showed, and these capabilities, can be applied in your business.

Tibor Vass (40:37):
Even if you start small, enabling these capabilities and automating a process-related task can create enormous savings in time and effort within a short period, without risking system operations or requiring a massive IT investment. These use cases can be added to your existing system, so do not worry if you have legacy systems or an older application that

Tibor Vass (41:14):
must remain in service. We are happy to integrate with all of them. We do not encourage you to replace your CRM, BPM, or other solutions, because we believe orchestration and integration of existing platforms are important to

Tibor Vass (41:32):
provide customer and agent experiences together. We help you move from on-premises solutions to the cloud, reintegrate those solutions in the cloud, or add new features to your system. Many customers select business-critical use cases, apply our system, and then grow by adding new processes and applications after the system proves itself.

Tibor Vass (42:05):
This is not a giant platform you have to buy before applying anything. Feel free to start with one or two use cases, then grow from there. These new AI capabilities can be applied in day-to-day practice, in the front office and back office, and for customers and agents alike.

Tibor Vass (42:35):
One additional capability we are proud of is integration inside a Genesys or contact center desktop, which is unique. CRM, BPM, and workflow vendors often want to retain their own desktop and do not allow their solutions to be placed inside another third-party omnichannel desktop.

Tibor Vass (43:03):
For us, it is not an issue. We can place our solution inside a dynamic iframe, which means you can use it without changing the customer or agent desktop, or even changing your applications. We can be embedded. Consider us an ecosystem and orchestration play. We are happy to provide further information. Thank you for watching this webinar. I hope you have questions, and we are happy to answer them.

Tibor Vass (43:32):
Please feel free to connect with us, watch a demo, or ask more questions. I will ensure that all of you receive the webinar recording and presentation slides. We will also attach a demo link showing how you can leverage these capabilities in real life. Thank you so much. Thank you, Max, for presenting. Thank you for participating, and I hope we will see you next time when we go even further into these use cases.

Tibor Vass (44:01):
Thank you. Have a nice day.