The idea that the future model of customer experience includes a high level of automation is already widespread. A whole ecosystem of developers and vendors is catering to an emerging mass market of bots and virtual assistants. And users and brands are discovering more applications of automated support over text and voice as bot libraries grow rapidly.

While the learning curve for users is a lot quicker than for command-line or form-based interfaces, and we all adapt to conversational interfaces, businesses have a window of opportunity to provide an efficient and effortless customer experience. In this window, technological capabilities and business processes will need to stop playing catch-up and start meeting rapidly evolving customer expectations.

Technical support in the IoT and connected home world

Most of the media coverage is on the use of virtual assistants in marketing, sales, or basic customer care. However, in technical support, the undercurrents of IoT and the connected home will drive an even quicker adoption and ROI. The multiple connected devices spreading around the home and the office increases potential set-up failures and interoperability issues, raises support costs, and (as OEM device margins decrease), lower support funds. The networked relationship between customers, products, and brands creates more complexity while simultaneously expanding to the early and late majority of the market. In turn, these user segments perceive product quality a lot more in terms of product experience; for example, the ability to self-install or receive instant support versus the actual technology. Combined, these factors aggressively force technical support down the virtual assistant and natural language path.

CSS Corp’s Yodaa: the Cotelligent virtual assistant

As a specialized technical support vendor, CSS Corp looked to address this reality with a new engagement platform with a common interface called Yodaa. Yodaa includes pre-packaged modules for premium tech support, analytics, virtual assistant, knowledge management, and mobile app, powered by machine learning and NLP.

CSS Corp developed Yodaa with three questions in mind:

• Is the platform intelligent enough to eliminate a call – i.e. can it identify a pattern among the registered issues and problems to proactively feed the product developer?
• Is the platform flexible enough to evolve – i.e. can it learn from the new issues that arise?
• Is the platform independent enough to remove the need for human intervention – i.e. can it create a fix by itself, for example with a return merchandise authorization, to identify the type of request and automatically trigger the process and push notification updates to the user?

These challenges might be part of a typical RPA integration: what CSS Corp adds is the substitution of human decision making with that of Yodaa. For example, when dispatching a new device, it goes through some 20 different checks such as validating the serial number, checking the warranty period and the bill of payment. Under a business process automation all these steps can be performed by a machine, but for support steps (e.g. in case of a dispute management), Yodaa uses natural language processing and text mining to remove human intervention, and applies machine learning to train itself to modify the business rules on a per case basis.

Over a four-month period this year, CSS Corp implemented the platform with a global gaming brand which consistently requires resolution of a high number of complex inquiries with a range of issues following a typical new version release. Before the release launch, agents created standard operating procedures within the platform’s drag and drop flowchart builder. After the launch, Yodaa learned from the live chat and email interactions how to (e.g.) access a new game level or reset settings to resolve gamers’ inquiries. So far, the vendor has been able to eliminate between 15-25% of the interactions associated with the new version release.

Still at the early stages of this engagement, CSS Corp expects Yodaa to deliver on average 50% reduction in call, email, and chat interactions for technical support, to improve agent productivity by up to 30%, and increase the customer satisfaction by 25%.

Bringing Yodaa to the call center with Amazon Connect

To jumpstart the adoption in the contact center, CSS Corp will integrate Yodaa with the newly launched Amazon Connect cloud-based contact center solution. Amazon Connect places phone, email, chat, messenger, and voice search interactions on a single CRM which enables the virtual agent to learn from one data source, answering simple queries and routing the complex ones to human agents. For clients with an existing knowledge base and standard operating procedures, Yodaa will search throughout for answers, but in the case of new or unsupported questions, it employs information retrieval techniques to obtain appropriate responses from the web. The ambition is for this continuous learning to eliminate and deflect an increasing number of questions, making human agents highly specialized subject matter experts. 

Technical support for the voice-enabled channels 

The next steps for the Yodaa developers are improving deep learning, and adding new business processes and verticals, e.g. pharma. Another focus area is support over the voice-enabled channels such as Amazon Echo, Google Home, and Apple HomePod. The company is working on adding skills and integrating NLP to resolve technical support questions addressed to voice engines. It has already set up demos where Alexa is guiding the user on how to set up a router or on the status of their device delivery.