When SAP Conversational AI Meets Robotic Process Automation

“Did you know that by 2020, analysts predict that you will have more daily conversations with bots than you will with your spouse?” Marie Myers, UiPath CFO.¹ 

Now that we’re in 2019, how does it make you feel?

But What Are Bots Anyway?

Imagine texting a number to get your replacement credit card after you accidentally lost it, and having it delivered without ever talking to a real human from the bank—that’s what bots are all about. To be precise, a bot is a function or an application that performs an automated task, such as setting a reminder, telling you the weather, or searching online. While these are easy and simple rule-based tasks, you can make them do more complicated tasks by adding intelligence to them.

For instance, if you add natural language processing (NLP) skills to the bot, then you can chat or better speak with them and have them respond to it. So instead of providing a fixed text (“Please replace my credit card with number XXXX-XXXX-XXXX-521″), you can have a more human-like conversation with the bot and use text like “…just like mine.”

Of course, I don’t have to explicitly explain that the conversation is going to be context specific. Meaning, when you say “Order a replacement card for me,” the bot understands who “me” is.

A Chat with an SAP Conversational AI Bot

Here is an example where I am going to chat with an SAP Conversational AI (Bot) and within three sentences, I’m going to get a replacement.

Did you observe what happened? A few things:

1. First of all, Bot understands English as a language (with grammar and other nuances)
2. Bot understands who YOU are, so it’s context specific
3. Bot has access to your order history. So not only does it recognize WHO you are, but it also KNOWS you (of course, within the digital spectrum)
4. Bot can process an image. (Yes!)

Deep Learning Adds Another Level to Conversational AI

Now let’s add a bit more complexity. Let’s say you want to replace a crystal figurine you bought for your wife. So instead of calling the company’s customer care, imagine they have a bot on their website that can not only talk to customers using NLP, but it also has deep learning skills enabled in it. (See the conversation below.)

What just happened—did you notice? We spoke with a “Robot.”

• Robot looked up all the information.
• Robot understood the image (using deep learning).
• Robot gave a recommendation using machine learning.
• And then finally, Robot performed a process automation (Yes, that’s RPA!) and ordered stuff for you. All while you were chatting/talking with it.

This is the next generation of bots—and this is called Intelligent Robotic Process Automation.

In my next blog, I will be talking about how we can actually build a bot. Furthermore, I will be talking about how to run the bots and using the concept of Blocky to optimize the bot.

Stay Tuned!

Learn More

• Read the rest of our Machine Learning Thursdays series blogs for more on artificial intelligence, deep learning, and predictive analytics topics.
• Visit the SAP What is Machine Learning? page and the SAP Conversational AI page
• Join us at:
  • SAPInsider BI, Analytics & HANA in Las Vegas March 19 -21
  • Gartner Data & Analytics Summit in Orlando March 18 – 21

¹ To Bot or Not to Bot – Is that the Only Question?