The following video has been created together with Raiffeisenbanken Service GmbH to show our joint journey in reaching best-in-class customer service in the financial sector.
What is smartCC?
An intelligent omni-channel assistant for customer service, which
- can “listen in” on calls, chats, email exchanges, exchanges on social media, etc.
- understands the ongoing conversation using advanced natural language understanding
- using a semantic model of the business domain and the related support cases
and can assist agents by:
- automatically identifying customer needs and relevant business cases
- proactively gathering information directly from the conversation
- providing guidance with the handling of a case
- presenting an integrated view of all relevant information
- taking over background tasks / interfacing with backend systems
smartCC can also connect to internal and external information sources to collect knowledge about the business, and make relevant information available to agents in summarized form in real time. It is also self-improving, as it learns “on the job” from implicit and explicit agent feedback during actual conversations with customers.
What makes smartCC different?
Thanks to our state of the art technology, smartCC:
- Performs real-time Natural Language Understanding (NLU) on a semantic level
- Can import and use any public or private domain ontology
- Can learn additional information about the domain from free-form documents, intranets, portals, etc.
- Enables coherent customer conversations that span multiple channels
- Can operate in real-time agent-support mode, or autonomously
- Self-improves while in use, thanks to the built-in AI
- Integrates well with different customer care center infrastructures and related tools and services
- Has a pluggable architecture that can support different commercial and open-source Natural Language Processing (NLP) components
Unique business advantages
- Businesses can capture their institutional knowledge and processes in ontology-backed business case models, using these as a potent basis for customer care services
- Its capability to understand and reason about the domain and about business cases means smartCC can handle complex tasks that require human-equivalent understanding of a situation
- Because of its basis in semantics, it doesn’t need pre-training to be effective — but it nevertheless learns and self-improves while in use
- Businesses can decide how to position it in the spectrum between live agent support and fully automated case processing, as it best fits their needs
”Our goal is to offer all our customers the same range and quality of our services on all relevant channels. Together with Contexity, we have therefore developed a completely new approach for modelling our business cases in a channel-independent and content-complete manner. This creates immediate benefits for all stakeholders. After a second phase, we will be able to partially or fully automate case handling, which will allow us to offer services e.g. around the clock, without waiting time, and with a quality that is only made possible with smartCC through its AI as well as its linguistic and semantic understanding.Andreas Putzinger // CTO // Raiffeisenbanken Service GmbH
Want to know how smartCC fits to your business needs?
smartCC in action
Through the combination of NLU and semantic business case modelling we reach a deep understanding of the business domain and processes involved and can therefore greatly facilitate the delivery of high quality customer services at high efficiency.
Explore the benefits of our approach by having a look at how smartCC works behind the scenes during a typical chat session.
Some more details
- Semantic modeling of business cases
- Real-time NLU for intent recognition
- Live agent support
- Omni-channel case handling
- Unique Customer and User Experience
- Continuously self-improving
Thanks to the use of ontologies and smartCC’s extensive semantic reasoning capabilities, businesses have full control over the system’s skillset, and internal experts can continue expanding and finetuning after initial deployment.
smartCC uses the definitions of business cases to decide what needs to be done once a case gets activated. For example, it uses the case models to understand communication with the customer, it retrieves relevant information from multiple sources, it automatically generates a user interface through which the agent can complete the tasks that need to be performed, and more.
With the smartCC tools, one can easily model business situations such as “Someone lost a bank card”.
Natural Language Understanding
smartCC uses a flexible NLP pipeline to process communication as it happens. This can be an ongoing phonecall, a text-based chat, an exchange over social media, an email thread, etc. The NLP analysis is followed by steps that determine the real semantics of what was said, and linking these to its domain knowledge, and to the business cases it knows about.
The way smartCC performs Natural Language understanding is based on concepts rather than on simple keywords. This means, for example that “I lost my credit card” and “My husband can’t find his VISA”, are both understood as the same type of situation, but with different persons being affected. Thus, it would know that, in the first case, it would need to identify what card of the caller was lost, while, in the second, it would know that the VISA-issued credit card of the caller’s spouse is what was lost.
With this level of understanding of business cases and real-world situations, smartCC can offer a multitude of types of support to agents:
- Automatic generation of form-based GUI to help guide case handling
- Assistance in the form of “prompts” the agent can use to ask for information
- Live observation of conversation and, with the help of the case models, identification of case-relevant information
- Pre-filling of form fields with extracted information
- Easy to use interactive controls that allow agents to intervene and choose from alternative possibilities, or enter information by themselves
- Automatic delivery of collected data to backend services on behalf of the agent — based on the modelled business case actions
Omni-channel case handling
smartCC keeps track of two levels of discourse for any open case:
- The first level is the current “dialogue” in a phone-call, chat, email, …
- The second level is the overall “conversation”, made up of all the “dialogues” that have happened to date
The benefit of this approach is that smartCC can provide a cohesive customer experience, always being aware of what information has already been exchanged, what questions are open, what are the next steps to take, etc. This makes the whole experience less frustrating for customers, who don’t need to repeatedly explain their case and the background, as well as faster and easier to process for agents.
Unique experience for the agent
… who receives real-time support and empowerment to handle customer requests, especially nowadays that most of contact center employees are working remotely
- smartCC helps to quickly identify customer needs and relevant business cases
- proactively gathers information directly from the conversation
- provides guidance with the handling of a case
- presents an integrated view of all relevant information
- takes over background tasks / interfacing with backend systems
Unique experience for the customer
… who now –more than ever before– try to resolve all their cases remotely:
- contact the business and interact using whichever channel is most convenient at the time
- no need to repeat any details, even in long and complicated multi-step processes
- faster service and easier case resolution
smartCC incorporates machine learning (ML) components which enable it to learn from what is happening while it is being used.
smartCC watches closely what the agents do: do they pick different / additional business cases to the ones suggested? do they correct data inferred by the system? do they ask new questions that haven’t been encountered before? The system then tries to understand what has worked fine and what hasn’t, in the context of the conversation, and use the additional knowledge to self-improve.
Through this learning-feedback-loop smartCC learns: different ways in which people talk about concepts in its knowledge base; how to better differentiate between similar situations that, nevertheless, require different handling; how to better assemble information split between multiple “dialogues” between a customer and different agents; etc.