I just returned from a quick trip to New Hampshire for a client engagement. Three or four weeks ago my client booked a hotel room on my behalf and I received an email confirmation from the Crowne Plaza. I went to put it on my calendar, and found that my Big Brother had already read my email and entered it for me. Aha! I thought. This time Brother Google made a mistake. Brother Google got the dates right, but he noted I was staying at the DoubleTree.
Turns out Brother was right. A week before my arrival the hotel changed ownership.
While the hotel was still sending confirmations out as Crowne Plaza in mid-April, as the contract with the Crowne Plaza did not end until April 26, Google had already data crunched the news and dates and got it right for me. Such is the nature of big data. It is a convenient, intriguing and scary all at the same time.
A number of traditionally consumer-oriented applications such as Google, Facebook and Amazon exploit the vast amount of data they collect about people, places and things along with artificial intelligence algorithms to make their applications stickier and ultimately drive up their revenues.
As consumers we have come to expect the applications we interact with to know many things about us and to be able to answer questions and help us accomplish tasks within a context that is meaningful specifically to us. For companies to be able to compete for our business, we will expect the same.
Forward-looking enterprises are starting to figure this out and are deploying artificial intelligence and bots both to improve customer service and reduce costs.
Today’s technology has come a long way from the early days of AI. Early attempts were script-driven applications that looked for keywords, applied values to these keywords, and responded accordingly. They were labor intensive to build and maintain and not very reliable, so applications were really just for curiosity sake, not of practical use. “DOCTOR” ELIZA is a well-known example of a robotic psychotherapist created in the 1960s.
Enterprises are deploying AI for a number of use cases including call center automation, marketing campaigns, customer self-services and internal employee processes. Anything that includes repetitive, deterministic tasks can be automated.
Pypestream is one up-and-coming company helping enterprises along this journey. I met with Pypestream earlier this spring at Enterprise Connect and was impressed with the list of large enterprise customers they are starting to build in various verticals including SlingTV, Solera, TriWest and Lincoln Center.
AI and bots were plentiful at Enterprise Connect this year, and with Natural Language Processing (NLP) and Natural Language Understanding (NLU) algorithms readily available via open source, vendors need to add value to differentiate themselves.
Pypestream uses a constellation of 5 NLUs, each designed to perform differently depending on the use case to deliver the best possible performance.
Pypestream talks about their “pragmatic approach to AI”. When I asked them what they mean by this I got a pragmatic answer. Start with automating repetitive processes, frequently asked questions, recurring payments, appointment scheduling, etc. and use the best tools for the job without overcomplicating it. In other words, if the question is “Would you like fries with that?” you don’t need a complicated workflow or machine learning scenario when the answer can be obtained through the use of two buttons: “yes” and “no”.
The platform has a variety of conversational interfaces for mobile, web, social and plain text end-user touch points. For example, it supports Facebook Messenger, with the ability to switch the session away from Facebook’s prying eyes mid-conversation to a secure connection.
Pypestream provides an end-to-end solution which has a number of advantages including providing a more secure solution and easier implementation. As far as implementation enterprises can choose three models: Pypestream can implement it for them as a turnkey solution, Pypestream can train the customer to build and deploy their own use cases, or the customer can go with a Pypestream certified partner.
Pypestream claims its customers are able to achieve payback within a year as well as 5X in savings compared to call center live agents and 30 percent call deflection in six weeks.
With this kind of rate of return it’s surprising more enterprises have not yet jumped on the bot bandwagon. Zvi Moshkoviz, Pypestream CMO, theorizes it is due to a lack of understanding about the technology and how it can apply to their business. “We find that once enterprises really understand what our technology is capable of doing, and that what we offer is much more than a ‘chatbot’, they clearly see that Pypestream will enable a solution to make life easier for their customers, “ Zvi said.
Perhaps in many cases it is just the old fashioned human feelings of fear of the unknown and unwillingness to take a risk. I wonder what ELIZA would have to say about that?
