Amazon Lex Automates Call Transcript Review for Chatbots

Amazon Lex Introduces Automated Chatbot Designer

During AWS re:Invent in Las Vegas, Amazon unveiled the Amazon Lex automated chat bot designer, currently in preview. This new feature streamlines the process of chatbot creation and training through increased automation.

Simplifying Chatbot Development

Swami Sivasubramanian, VP of Amazon AI, announced the new capability, stating it can reduce chatbot design time from weeks to mere hours. This acceleration is achieved by leveraging sophisticated natural language understanding powered by deep learning.

Developers can now rapidly establish a foundational chatbot utilizing historical call transcripts with just a few clicks. The system intelligently analyzes these transcripts to define user intentions.

Intent Recognition and Analysis

The automated chat bot designer can typically process 10,000 lines of transcripts in approximately two hours. It identifies key intents, such as “file a new claim” or “check claim status.”

Crucially, the tool ensures clear separation between these intents, minimizing overlap and eliminating the need for iterative trial-and-error adjustments.

Addressing the Challenges of Manual Chatbot Design

Without this automation, chatbot design is a laborious and time-consuming task for developers. Organizing a chatbot effectively is complex and susceptible to errors.

It demands a deep understanding of spoken language and human interaction. Without specialized expertise, developers can spend countless hours analyzing historical call transcripts to pinpoint common user requests and the necessary information for resolution.

The Broad Applications of Chatbots

Chatbots represent a common application of artificial intelligence. They can be deployed internally to address employee inquiries regarding topics like computer ordering or health insurance enrollment.

Alternatively, they can serve as a customer service interface, gathering essential information and resolving simple issues, while escalating more complex inquiries to human agents.

Amazon's Position in the Chatbot Market

While numerous startups are focused on improving chatbot accuracy, Amazon’s offering presents a compelling solution for customers already utilizing its AI and machine learning services. This represents a logical extension of their existing platform.

Availability and Pricing

The Amazon Lex automated chat builder is available in preview as of today. During the preview period, developers can utilize the feature without charge.

Upon general availability, usage will be billed based on the time required to analyze transcripts and identify user intents.