By Pat Calhoun, CEO and founder at Espressive

NLP is the Next Buzzword

In 2019, artificial intelligence (AI) was thrown around as a requirement for chatbots. Unfortunately, the specific requirements of AI were not clear, causing confusion when evaluating options. This ultimately caused AI to become a "checklist item" for chatbot RFPs, rather than a technology that could be leveraged to provide business value.  

In 2020, the same thing will happen with natural language processing (NLP): NLP enables chatbots to process and analyze natural language data. However, improving the chatbots ability to respond with accuracy requires a large amount of data. In addition, for a self-help chatbot to be successful with employees, it must have a consumer (think Amazon's Alexa)-like approach, as well as the ability to teach and update content with ease.

Without a clear understanding of these NLP requirements, chatbot initiatives will continue to fail. 

There is no shortage of NLP and chatbot toolkits on the market. Here are a few NLP toolkit requirements to look for:

• NLP Accuracy: Consider how Amazon data scientists have been able to improve Alexa's accuracy thanks to the millions of consumers using and speaking to Alexa every day. This is no different in the enterprise world: if your chatbot is to be successful, it must come with an NLP stack pre-built with enough employee language data to effectively understand all possible permutations of any phrase.
  

• Teaching Content: An NLP stack must enable you to add content quickly in order to adapt. No company is ever stagnant; you must have the ability to simply enter a phrase and add an answer without a lengthy training process.
  

• Scaled to the Enterprise: The ability to understand employee language across the enterprise means your chatbot can respond to multiple teams within IT and beyond (e.g., HR, facilities). However, to effectively scale, content management must be democratized by empowering subject matter experts across departments to easily manage their own content.
  

• Simple and Safe Testing: With the nuances of language, one simple change to NLP content can have a negative impact on what appears to be a completely unrelated phrase. To avoid this, a tool should make it simple to capture thousands of phrases and run regression testing at any time. An NLP stack must also provide a way to make and test changes in a safe development environment, versus a live production environment, before pushing them into production.
  

By understanding these requirements, NLP will become more than a buzzword for chatbot RFPs. Instead, it will give self-help chatbots a consumer-like experience that can be scaled throughout the enterprise. Only then will employees return to the chatbot for help.

##

About the Author

Pat Calhoun is a visionary leader with an intense focus on user experience and customer adoption. He has founded two companies - Airespace and Espressive. As CEO at Espressive, Pat is set to transform the enterprise self-service experience to a consumer-like approach that drives employee adoption and significantly reduces help desk calls. Pat's first startup, Airespace, grew revenues to over $80M in two years before being sold to Cisco for $450M. 

Most recently, Pat served as senior vice president of product at ServiceNow where he was responsible for ServiceNow applications. Prior to that, he was general manager of the McAfee network security business. Pat also served as both CTO for the Cisco $14B switching, routing, wireless, and security access business and GM of the Cisco identity business. Pat holds 35 patents and has been published in more than 16 publications.