Traditional Chatbots Vs Kate: What Makes Emplay’s Knowledge Engine Special?

April 12, 2022

Traditional Chatbots vs Kate: What Makes Emplay’s Knowledge Engine Special?

Chatbots are widely accepted and used in various industries for multiple use cases. From automated customer support to order confirmation, and from internal helpdesk support to information hunting, chatbots promised to take organizations to the world of autonomous support.

However, there’s something about chatbots that makes them not very intelligent. For one, they lack context, meaning they can’t understand the meaning of an ongoing conversation to give the best results. Second, they are trained manually, which is a massive setback in terms of scalability and flexibility.  

In this article, we dive deeper into the limitations of traditional chatbots and point out what makes Kate — Emplay’s proprietary Knowledge Assistant and Training Engine — truly next-level.  

Traditional chatbots — the problem lies in the way they’re trained  

In simpler words, here’s how a traditional chatbot is trained.  

1. Define specific use cases the chatbot can handle e.g. HR Benefits Assistance

2.Create a database of questions and answers (FAQs) with the help of SMEs. Say 100 FAQs.

3. Generate a list of alternate utterances users can use to ask the same questions. Now 500 utterances.

4. Feed the FAQs to a bot and add alternate utterances to an NLU engine.  

5. Keep training the chatbot forever.  

It’s a tough job keeping your chatbot up-to-date due to the manual, time-consuming process required for it.  

Another problem in traditional chatbots is the lack of context. Even though most chatbots claim to be “intelligent” and “as close to humans as they could” — these promises don’t translate to a meaningful experience.

To be intelligent throughout a conversation, a chatbot has to understand the context and provide answers related to the subject in question and maybe ask qualifying questions when the context is ambiguous. Most chatbots fail miserably at that.

To help you understand the meaning of “lack of context”, here’s a quick example conversation between a human and a chatbot.  

Human: “What is my insurance premium?”  

Bot: “$2800”

Human: “When is it due?”

Bot: “I’m afraid I do not have an answer for that”

You see, the bot did not understand what was “it” referring to.

Also, the bot did not check if the question was about health insurance, dental insurance, or vision care.

Kate by Emplay — what is all the fuss about?  

Enter Kate — the first self-trained knowledge assistant.

There are multiple ways Kate is different from a traditional chatbot. Here’s a breakdown of what makes Kate special.

1. Self-trained; no need to write FAQs and train for language

First, you don’t need to engage SMEs to convert knowledge into FAQs and then generate alternate utterances to factor for natural language questions.

The only thing you have to do is to connect KATE to your data sources. The content could be in any format, web links, documents, or videos. KATE, automatically converts content into FAQs and conversation flows. When the user asks a question KATE extracts answers from deep within documents and presents the results. Furthermore, it redirects the user to the exact page, slide, or video segment where the answer lies.

2. Contextual conversations

Unlike most other chatbots out there, Kate recognizes and understands the context during the conversation. KATE’s context engine builds the context based on who the user is, what is she doing, prior questions she asked, and similar questions similar user profiles usually ask.

But that’s not all, Kate takes it a step further. It suggests automated follow-up questions if the context is ambiguous. For instance, when a user asks, "How to claim expenses"? KATE asks qualifying questions about what kind of expense the user has in mind. KATE is able to do this without being manually trained.

3. Continued Conversation

KATE is designed not just as an answering machine but as a problem-solving machine.

Using its intent identification engine KATE is able to understand whether the intent of the question is informational, navigational, transactional, or commercial to provide the right answer but recommend the next question or action the user may be interested in next.

E.g., if the user asks about the maternity leave policy KATE can respond with informational content as well as recommend that the user read about maternity benefits, maternity care, and how to claim hospital expenses.

Thanks to the power of AI, Kate can dynamically sequence the questions based on similar questions asked in the past by similar users and related topics.

The bottom line

Chatbots have a huge potential to provide conversational convenience and process efficiency. However, their technical limitations are impacting their performance and reputation.

KATE reshapes the chatbot and knowledge discovery market and finally fulfills the promises conversational AI technologies originally made.