Chatbots, Continuity, and a Helping Hand


Aspect’s acquisition of Voxeo several years ago was largely motivated by the success and innovation of their long-standing “design once, deploy anywhere” concept that gives customers the freedom to add voice and text self-service channels with minimal effort. Combine that with Aspect’s history of delivering agent-assisted interaction management and WFO solutions in the contact center and we have something special in terms of guaranteeing seamless omni-channel connections with consumers.

Beautiful asian businesswoman at hotel room

We can deliver a high level of autonomy for customers to solve simple problems on their own, thereby setting the stage to elevate the role of contact center agents to trusted advisors in solving more complex issues. We then added natural language understanding (NLU) to our portfolio last year so we could unlock another powerful capability for empowering consumers to get answers to questions. (Have you met Edward yet?)

Knocking down silos has been the guiding inspiration to these efforts. “Design once, deploy anywhere” in and of itself allows you to deliver a consistent customer experience on all channels — but the ability to deliver a continuous, context-aware customer experience across those channels, even between self-service and live agents, that is what defines a real customer service solution.

The deployment of “chatbots” siloed from the larger customer service landscape has had its share of challenges. Many of the first generation of chatbots have been labeled “frustrating and useless” (Gizmodo), “spammy” (The Guardian) and “buggy at best” (Tech Times). The promise of automation was greater than the actual implementation.

It turns out many of our favorite automated personal assistants that provide more seamless experiences than last month’s wave of chatbots have humans behind the curtain. Many concierge services are using supervised learning, hiring a multitude of AI trainers to scrutinize any data where the AI’s confidence in its assessment is low, and adjust it accordingly. This adjusted data can train the AI model to answer similar questions better in the future; in the short term, however, the customer won’t be aware of whether it was the AI or adjustments by a person that yielded the right answer to any particular question.

For customer service, where a personal touch is often appreciated, our approach is to automate the most common inquiries and support a transparent transfer to a live agent – who can identify as such – when needed. Depending on the nature of the inquiries fielded by the live agent, the information gathered from these interactions that make it to live status could be used to give additional training to their bot, or can remain within the domain of live agents.

Aspect’s chatbot Edward likes to remind Radisson Blu Edwardian’s customers that he’s a neophyte – eager to help, but still needs to learn. He’ll eagerly bring in a person when he knows they can do a better job, giving them context and then moving out of the way so that they can have a real person-to-person interaction. Though this arena is still new, we’ve seen that in the market, fully automated service can result in customer dissatisfaction and backlash. Automation that can, when needed, segue to live assistance with full transparency and continuity provides the most seamless and successful experience for customers. That’s our story and we’re sticking to it.

One thought on “Chatbots, Continuity, and a Helping Hand

  1. I ⅼike what you guhѕ are usuаlly up too. Sսch clever
    work and ϲoveгage! Keep up the fantastic works guys Ӏ’ve included
    you guys to my blogroll.

Comments are closed.