Customer-Facing Chatbots
This page contains a description of some of the customer-facing chatbot solutions we at Kmbara provide. Our chatbot solutions use advanced technology, machine learning and statistical models to enable your customers and prospects to get the answers they need, quickly and easily.
The Challenge
Every business wants to be able to communicate with their customers and prospects, but resources are limited: trained call center employees and customer service representatives are costly and can strain thin resources. Some of the questions that customers and prospects have seem to require trained professionals to deal with, but most are relatively simple and are asked frequently.
The traditional solution to this challenge is to create a Frequently Asked Questions (FAQ) page on a company website. This can help, but it's not a perfect solution for several reasons. First, customers may not be willing to make the effort to find and closely examine an FAQ page, especially one that's quite long. Second, an FAQ page does not customize itself for the particular needs of each unique customer. Third, an FAQ page is "low-tech" and lacks the sophistication of a chatbot that can impress prospects.
We at Kmbara can create simple chatbots that are trained to intelligently answer FAQ's and run 24/7 to talk to customers and prospects all around the world. We can create them quickly and at low cost. We can also train chatbots to be more intelligent than a simple FAQ page, including personalization for each customer and the ability to answer more advanced, complex questions.
Our Solutions
Our chatbot solutions use advanced software tools and machine learning methods to intelligently communicate with customers and prospects. Here, we'll present some details about several of the most common solutions.
FAQ Bot
The simplest chatbot we create is an FAQ Bot, that can be quickly trained to answer a company's FAQ's.
If your FAQ's are already written, we can quickly configure one of our existing chatbot templates with your FAQ's to answer the questions. This configuration requires converting natural language text to a numeric form. This is a common requirement for natural language processing (NLP) methods, since human language is messy and complicated, and numbers are ordered better and easier to work with using quantitative data science and machine learning techniques. Converting text to numbers can happen in one of several different ways. One of the conversion methods converts words to vectors in a format called "TF-IDF", which is short for term frequency-inverse document frequency. The TF-IDF format is easy to calculate quickly, and for each of the FAQ's, it takes into account how frequently each word appears, and how unique it is in each of the questions.
After converting text to numeric vectors, we proceed with a step called topic modeling. This step can also proceed in several ways. In one of the simplest methods, we calculate a "similarity score" between user inputs to the chatbot, and each of the existing FAQ vectors (in TF-IDF format). The existing FAQ question/answer combination that is the most similar to the user input is returned as the topic that the user is interested in.
Every chatbot will need a user interface. For this part, we rely on our top web developers, who can create chatbot web applications and widgets using a variety of frameworks. One simple approach is to create a simple Flask application that enables website visitors to type in their questions and get the bot's response. The particular solution we use for the interface will depend on your existing site and your particular needs, and we're flexible enough to accommodate any needs.
The final FAQ Bot will face customers and prospects from around the world, 24/7, and intelligently respond to the most frequent questions that users have. It will be flexible with user inputs, able to understand questions asked in several different ways, and will be able to accommodate slang, misspellings, and rephrasings of all kinds.
More Advanced Chatbots
FAQ Bots are like the first step in an evolutionary process that culminates in advanced chatbots that can answer a huge variety of questions in multiple languages, and perform complex multi-system reasoning to provide an accurate answer.
Enhancing a simple chatbot to make it more advanced can happen in several ways. One way is to "teach" the chatbot other languages: an "English-speaking" FAQ Bot can learn Spanish in a few minutes, for example, as long as the FAQ's have already been translated. Adding support for other languages can have some complications, including support for different character sets, but we can overcome all of those challenges.
Another enhancement that can improve a chatbot is to connect it to other enterprise resources. For example, a simple FAQ bot could answer a question about data by providing a canned answer or URL suggesting where the data could be found. But an advanced chatbot could have the capability to convert user questions to SQL queries that could then connect to enterprise databases to actually pull the data in question and provide it to users.
The most advanced chatbots can be configured to perform quantitative analyses, for example by making strategic recommendations about budget outlays, or performing optimization for operations scenarios.
Other enhancements that can improve chatbot performance could have to do with the user interface. We could configure a bot that answers SMS messages or emails, or even one that understands voice commands and talks back.
Quick, Accurate, Dynamic Solutions
Chatbot technology has progressed by leaps and bounds in recent years. For experts like us at Kmbara, a basic chatbot solution can be configured in a matter of no matter than a few hours. We can make chatbots quickly - and we can make them highly accurate. The initial setup isn't the end of what we can do: we can continue with you throughout a development process where at each step we'll increase the chatbot's capabilities. Like a child growing up or a species evolving, we create dynamic chatbots that are always improving. Contact us now to find out more about what we can do for you.
