In 2011, Apple introduced Siri, a personal virtual assistant, on its smartphones. This was the world’s first AI-enabled chatbot incorporated into a mass-produced merchandise. In 2014, Microsoft introduced its version of a virtual assistant, Cortana. The same year, Amazon launched Alexa (Echo smart speaker) that has since enjoyed significant commercial success. Then in 2016, Google launched its Google Assistant. Today, AI chatbots and virtual assistants are everywhere – accessible via mobile apps and websites, over phone lines, and on smart speakers and physical robots. In coming years, broader and deeper applications using chatbots or virtual assistants can be anticipated. In other words, the era of “conversational AI” has already begun.
The terms “chatbot” and “virtual assistant” are often used interchangeably. In essence, a chatbot is an AI-based software that can autonomously answer its user’s inquiries. A virtual assistant not only answers questions, but also carries out specific tasks or workflows according to its user’s instructions. Based on their usage scenarios, they can be designed for personal, home, or enterprise use, and the formats of the human-chatbot conversation can be vocal, textual, or both.
A successful enterprise-grade chatbot (hereinafter referred interchangeably with virtual assistant) can help reduce HR costs as well as enhance customer satisfaction in areas such as marketing, sales, and customer service; it can also assist an enterprise in automating internal and external workflows. Therefore, an enterprise chatbot can contribute to both the enterprise’s cost and profit centers.
Every enterprise chatbot is unique – it handles information and workflows specific to the enterprise it serves. The capabilities of an enterprise chatbot can also be gradually added over time instead of all at once. Since there are no identical enterprises – even among those in the same field, customization of various degrees is necessary for the creation and evolution of an effective enterprise chatbot.
Due to differences in their underlying AI technologies, expandability, and customization flexibility, the capability and quality of chatbots available on the market can vary substantially. Their common benefits and drawbacks are listed as follows.
Common benefits offered by chatbots:
- They can provide 24/7 services.
- They can offer immediate, professional, and honest responses when dealing with relatively simple or routine inquiries.
- When used in tandem with enterprise websites or public messaging platforms (such as LINE, WeChat, or Facebook Messenger), they can deliver interactive user services via these highly accessible channels.
Common problems suffered by chatbots:
- They often misunderstand the nuances in human languages.
- They are often unaware of the context of a conversation.
- It can be difficult for them to handle complex conversations or workflows.
- When a user encounters difficulties communicating with a chatbot, there is often no adequate remedy, resulting in poor user experience.
- A general-purpose chatbot cannot fulfill the needs of an enterprise, but an enterprise-specific chatbot often comes with a high cost of customization.
Aiming to solve the above problems, D8AI Inc. has developed two proprietary software engines, based on deep learning (BERT and its derivatives) and machine learning algorithms, for producing and training custom chatbots: a natural language understanding (NLU) engine and a contextual dialog management (CDM) engine. Furthermore, to fulfill the need for an enterprise to continue updating and enhancing its chatbot after its initial launching, we also provide a comprehensive software platform for the enterprise to analyze and manage user-chatbot dialogues. This platform is designed such that a single administrator (no technical background required) can easily train, test, and update the chatbot model.
As it serves customers and enterprise employees, an enterprise chatbot also collects valuable data for the enterprise (for example, inquiries from customers, logs of workflows, etc.). Such data are ideal for business analysis, discovery, and prediction. Therefore, for many enterprises, designing and deploying an enterprise chatbot can be a powerful first step towards AI empowerment.