VANDA is a digital customer assistant that is capable of performing dialogue-based communication with a client both on chat platforms and over the phone.

VANDA can automate different scenarios while communicating with the client in natural language dialogue:

  • Determining the client’s intent to direct him or her to the appropriate group of human agents
  • Providing answer for the clients’ problems using static knowledge base
  • Guiding the clients through complex business processes to retrieve information dynamically or to initiate transactions based on negotiated parameters

Appointment scheduling, balance enquiry, conference registration, limit modification, submit electric or gas meter reading, problem reporting, solution searching, satisfaction survey… It is simpler with VANDA than with an ordinary IVR. We even let VANDA handle the beer tap for our guests at T-Systems Hungary events.

The buzz is around everywhere: virtual assistants are coming and replacing humans in many areas. Digital personal assistants (Alexa, Siri, Cortana etc.) supposed to have general intelligence and to learn freely based on their ongoing conversations.

In the enterprise environment, however the digital assistants should have business specific knowledge and should direct the clients towards well-defined business processes via guided dialogues. VANDA is this kind of trained assistant who maintains but not pushes its persona over the conversation and who is trying to provide a personalized and effective service for the client.


Intelligence is a rather complex phenomenon. An intelligent software solution might realize a few or many aspects of it. This is the case with VANDA as well:

  • VANDA has a decent language intelligence with strong semantics interpretation. She recognizes written and spoken language with very good accuracy independent of individual differences (80+ languages and dialects are supported as of today).
  • She has a limited general knowledge about the world and about herself that makes dialogues more human like.
  • She can be trained to handle business related processes using samples or by manual configuration.
  • She can gain information from various sources as databases, FAQs and other data sources.
  • She can learn about the history and behavior of her partner and that can form her reactions.
  • She is intelligent enough to realize when she reaches her limits and offers other form of help.


  • Capable of handling 86 languages/dialects;
  • Can be used both in voice and chat media using the same technology and language configuration;
  • Can be accessed from different applications such as phone, web chat, Skype, Facebook Messenger, Viber;
  • Capable of customizations and integration to complex business processes and back-end systems;
  • Capable of adapting to the personal attributes of individual clients (personalized assistance);
  • Can utilize biometric authentication features;
  • Provides detailed reporting to support business process (dialogue) fine-tuning and for business analysis;
  • Can be implemented on-premises or be provided from cloud;
  • Several payment schemes available: either up-front or on a monthly basis”.


VANDA is built on top of Nuance’s leading voice recognition, text to speech, language semantic and voice biometrics modules extended by T-Systems Hungary’s software components that helps building personalized, context based dialogues and that integrate with legacy systems.



The very same logic and configuration can be used both on phone and on chat platforms.


VANDA can authenticate customers on the phone in various ways:

  • Recognizing one’s voice while saying a standard sentence like “My voice is my password”. She is good enough to recognize you even if you have a bad cold. Moreover, she can detect if one has “stolen” your voice and tries cheating by playing back your recorded voice.
  • Recognizing one’s voice after having listened to several sentences during the conversation.
  • In addition, if this does not work, she might still ask you to type in your PIN using the phone buttons.

During a chat session, VANDA uses some other authentication methods:

  • She will ask you to provide your PIN code or password showing standard hidden typing: *******
  • Whenever more security is required, she will send a one-time password to your mobile phone and request you to use that one instead of a regular PIN code.


VANDA should know about your business processes and about your data to be able to handle your client’s requests intelligently. Therefore, careful design and most probably a certain level of integration will be necessary to let VANDA start working for you.

  • If you have an existing IVR self-service application or a web based FAQ, it could serve as a first step in describing your processes and data. In case you are just willing to implement self-service for your clients, your processes should be defined carefully.
  • One does not have to train VANDA to understand spoken language. She already speaks 80+ languages. However, she should be trained about how your clients are expressing their intents. This requires quite a number of sample sentences from client dialogues. You should consider where this data might come from.
  • A “just a simple proof of concept” will not work. A dumb virtual digital assistant will not be a good proof. Check our demonstration instead if you are interested in VANDA’s capabilities. However, VANDA might cover only a part of your processes at the beginning and later extend it. Nevertheless, one should implement those few processes with VANDA fully, not as a pilot.

Ask for our free demo to hear how VANDA can help you!