Without dominant vendors seizing the lion’s share of the conversational AI market, many view it as a vast blue ocean of opportunity.
Yet, since the start of the COVID-19 pandemic, many more ships now sail those waters.
Indeed, the space has become congested, and few vendors stand out from the crowd.
AWS is one of those that does, with its Amazon Lex solution that sits inside its Connect platform.
As such, Lex has become widely used within contact centers, as AWS leads the CCaaS market in customer acquisition with Connect.
Yet, it has not only become a prominent conversational AI platform because of its accessibility. Indeed, analysts generally consider Lex as a market-leading platform – as recent Constellation Research suggests.
One of the features within the platform that perhaps underlines this status is its automated chatbot designer. Alone, it differentiates Lex from its competitors in the vast blue ocean.
How? By automating much of the traditional toil of designing best-in-class virtual agents.
Find out how it does so after considering all the conventional challenges of bot building.
The Traditional Toil of Bot Building
The first stage of bot building involves isolating customer demand drivers. Without an analytics system, this hard graft. After all, contact center disposition data is often inaccurate.
Moreover, it is often tied to IVR inputs, which managers rarely review.
As such, bot developers typically spend weeks trawling through conversation recordings and transcripts to identify the most pressing reasons why customers contact them. These are often referred to as "intents."
The process is not only time-consuming and error-prone but lacks efficiency, as the contact center struggles to accurately quantify the prominence of each intent.
As a result, it is extremely difficult to quantify ROI for the bot.
Furthermore, some intents may overlap or go missing, confusing the design process further and ultimately leading to a frustrating customer experience.
Such inefficiencies also make it tricky to identify the customer’s desired outcome for each intent and build a logical flow from the initial query to that outcome.
Indeed, the entire process is error-prone, and perhaps a significant reason why 60 percent of customers still face frequent disappointment in their chatbot experiences.
Amazon Lex Automated Chatbot Designer Cuts Through the Toil
Last year, AWS released its Amazon Lex Automated Chatbot Designer. "It uses machine learning to automatically design a chatbot in hours, instead of weeks," said Annie Weinberger, Head of Business Applications Product Marketing at AWS in conversation with CX Today.
Developers can start by uploading their transcripts into Lex, where the chatbot designer analyzes those transcripts – using machine learning – and creates an initial chatbot design that can be reviewed and deployed.
"It can analyze thousands of lines from transcripts within a couple of hours. This reduces developer effort and the time it takes to create a chatbot. But, because of those ML-powered intents, it leads to a better customer experience."
Alongside this base design, the solution churns out associated phrases and a list of the information required to resolve each intent. That may include customer, order, or policy numbers.
Harnessing this extra information, developers can iterate on the design and ensure the necessary integrations are in place to retrieve the data.
From there, teams may brief senior agents and leverage their expertise to adjust bot responses, ensuring a smooth experience.
Finally, the business can test, tweak, and deploy the conversational AI solution.
