Avatar

Chatbots are a rising support model. With the advent of artificial intelligence and machine learning, chatbots can often provide resolutions to issues faster and more accurately than human support agents. According to Business Insider, 40% of internet users worldwide prefer interacting with chatbots instead of human agents virtually. By 2024, Insider Intelligence predicts that spending on chatbots worldwide for consumer retail alone will reach $142 billion—compared to $2.8 billion in 2019.  It is now common to find chatbots on retail, healthcare, and corporate websites and mobile apps. Chatbots can be designed to provide different types of expertise and functionality. When properly trained, they are very good at providing answers to specific queries. They automate common tasks, reduce time to service, and increase efficiency and customer satisfaction.  

If a person cannot tell whether a response is coming from a machine or a human, then a bot has passed the Turing Test. The biggest challenge for chatbots has been understanding the nuances of human language, especially technical jargon.  With advancements in machine learning and deep learning with transformative algorithms like Bidirectional Encoder Representations from Transformers (BERT), chatbots now sometimes perform better than many humans in understanding written intent.  It’s also easier than ever to create chatbots, with minimal programming necessary. Simply provide examples of interactions and a bot learns for itself how to respond.  

The Cisco Networking Chatbot   

The Cisco Enterprise Networking team foresaw a need for a chatbot to streamline support work and make it more efficient.  The Cisco Networking Bot (cnBOT)  is designed to empower internal support personnel, customers, guests, and partners by providing digitized Cisco product information in an intuitive way. cnBOT has experienced overwhelming growth, with thousands of users and 10s of thousands of queries.  

The cnBOT is available on the web, on product support pages, via the cnBOT Webex Team Space, and via the Cocoa Bot interface. cnBOT is also supported by analytical modules and helps generate leads for product sales and migration services.  

The cnBot (Figure 1) is a cloud-ready product that uses microservices built with scalability in mind. Its features can be prototyped quickly and independently, and its reliability is unmatched because of its distributed processing architecture. As a native cloud application, the cnBOT has built-in resiliency.  

Figure 1. Cisco Networking BOT Features
Figure 1. Cisco Networking BOT Features

 

For an example of the cnBOT in action, a request for information about Cisco Catalyst 9000 Series software compatibility generates a link to the matrix shown in Figure 2 and other related links. 

Figure 2. Cisco Networking Chatbot Use Case: Cisco Catalyst 9000 Compatibility
Figure 2. Cisco Networking Chatbot Use Case: Cisco Catalyst 9000 Compatibility

Another example of the cnBOT in action is in response to a query about migrating from the Cisco Prime Infrastructure (PI) to Cisco DNA Center. A high-level migration task list is displayed (Figure 3) with a prompt at the end to do a self-guided migration. 

Figure 3. Cisco Networking Chatbot Use Case: Migration from Cisco Prime to Cisco DNA Center
Figure 3. Cisco Networking Chatbot Use Case: Migration from Cisco Prime to Cisco DNA Center

Future Development 

The cnBOT started as an idea to facilitate Tier 2 workflow automation, reduce time to get critical information and provide a tighter integration loop with stakeholders — although the opportunity and scope can be much broader. Diverse use cases and different audiences can be served by the cnBOT.  We are working with multiple groups at Cisco to integrate their workflows into the cnBOT and looking for ways the chatbot can provide services through other bots at Cisco, including the CX Cloud bot, so customers can be provided with seamless support experiences regardless of the questions they ask.    

The cnBOT can be used for other types of communications, including voice input and output, like Apple’s Siri. It can also provide collaborate services like push notifications. Further investment will enable research into how customers use the product and how to mitigate pain points in their interactions. Longer-term, we anticipate developing many additional services for the cnBOT, including offering a standardized platform for easy integration.  

Chatbots are here to stay and will get smarter.  Bots in gaming have beat some of the best human players in the world with just a few hours of training. Machines are getting much better at understanding human language (sometimes better and faster than humans). More interactions with bots await us all, for greater efficiency and faster service that improves customer confidence in Cisco products and services. 

 

Check out Cisco Networking Bot to experience it firsthand and give your feedback at cnb@cisco.com



Authors

Prasad Chebrolu

Vice President, Products & Support

Intent-Based Networking Group