How to Use AI to Serve Your Customers Better

Add Your Comments

Among the early benefits of artificial intelligence systems, machine learning and natural language processing have the potential to transform customer service processes that are too often frustrating for customers and expensive for web hosts and service providers. A number of tools are currently available to enable companies to overhaul their interactions with customers contacting them for help by telephone or over the internet.

Customer service or sales calls often require that customers listen to menus of options and provide information–sometimes more than once–before a company representative even begins helping them. This experience can cause frustration, even when successful in the end.

Advanced automation services leveraging AI provide an alternative approach. For example, chat and messaging services like Facebook Messenger and Kik offer chatbots, which simulate human conversation through AI. More than 100,000 chatbots were created in their first year of availability on Messenger, according to VentureBeat. A survey by Oracle found that 80 percent of companies will be using chatbots by 2020, Business Insider reports.

Below are some of the options web hosts and service providers can use to enhance customer service experiences, without making massive investments in call centers and teams of expert representatives. Each option emphasizes its delivery of real-time analytics, easy-to-use data dashboards, and support for multiple languages.

Amazon

The Amazon Lex chatbot platform integration with Amazon Connect reached general availability at the end of June. This integration allows Lex chatbots, which use the same conversational technology as Amazon Alexa, to connect with customer databases and software applications to perform tasks. Lex passes issues it is unable to resolve to human representatives along with full customer information, and the system keeps interactions consistent to avoid errors.

IBM

Customer engagement is also one of the skills IBM has had Watson learn, to power Watson Virtual Agent. It includes pre-trained industry and domain knowledge, an engagement metrics dashboard, and deep analytics capabilities. Watson Virtual Agent is available for a free 30-day trial, and, like all of Big Blue’s products, it is supported by a huge community and library of docs.

Live!Zilla

Live!Zilla is a live help and support system that monitors website visitors in real time, based on open source software. It is sold as a perpetual license, rather than a subscription, and it protects privacy by running without cookies and with masked IP addresses. Live!Zilla integrates with any website and offers support for more than 30 languages.

Sales Syntax

Sales Syntax (formerly known as Crafty Syntax) was recently updated from a fairly basic live help software to include customer relationship management (CRM) tools such as referrer and page tracking. It allows for multiple operators, multiple departments and multiple websites. Its open source code is fully customizable, and it is offered with free, “unbranded” and “professional” versions.

Nanorep

Nanorep provides virtual assistant and chatbot solutions, collects and communicates customer information for live agents, and supplies real-time customer experience analytics. It uses patented natural language processing (NLP) technology boasts over 500 deployments. Nanorep touts the seamless transitions it provides customers with its smart channeling, and says it supports “almost any language.” It can also provide a 50 percent increase in customer satisfaction, according to the company, with a 45 percent reduction in contact center volume.

Twyla

Twyla is a mobile-focused AI chatbot provider. It works on a variety of channels, including Facebook Messenger, Amazon Alexa, Google Assistant, WeChat and Telegram, as well as over the web. Its AI learns from uploaded chat logs, and it can be configured to provide the best options, answers and conversational tone.

Pandorabots

Pandorabots, which has more than 225,000 registered developers, provides API access to SDKs, a hosting platform for text or voice chatbots, and Playground, a free development environment. Pandorabots both builds bots and provides support for companies building their own. Bots built with the platform have won the Loebner Prize for human-like AI.

Ada

Another recent startup is Ada, which claims chatbots can be set up in minutes without requiring a developer. The company offers robust AI that asks clarifying questions, and says it reduces support volume by 20 percent or more in one week. Ada can be deployed to popular platforms like Facebook Messenger and Slack, and offers a free 14-day trial.

Build-Your-Own Chatbots

Service providers considering AI for automated customer service can also build the functionality themselves. Build-your-own chatbot efforts can be supported by tools from Google, such as its chatbot analytics platform Chatbase, or bot-building software like ChatFlow. The maker of ChatFlow, Kitt.ai, was acquired by Chinese search giant Baidu in early July. Microsoft offers a Bot Framework for Node.js. ChatScript, an open-source tool written in C and C++, is another Loebner Prize winner. The engine is available on SourceForge, and has been evolving since 2009. Facebook offers a similar easy bot-making platform for Messenger, and several third-party companies offer bot-making templates for Facebook and other social media platforms.

Looking Ahead

While skeptics certainly remain, rapid advances in AI and NLP are constantly making customer service bots more effective. The market is still developing, with many companies currently operating beyond those mentioned above and many new entrants joining.

Companies providing AI customer service bots claim big reductions in time to resolution, agent workload and cost, as well as increases in customer satisfaction. Given the customer support demands on web hosting and service providers, and the recent maturation of AI and the chatbot market, those that have not yet deployed automated customer service to common touch-points should consider doing so.

Add Your Comments

  • (will not be published)