Building the complete virtual assistant with Watson Assistant

By | 4 minute read | May 5, 2020

We are now well into 2020, and the world continues to face a once-in-a-century global pandemic. The COVID-19 coronavirus pandemic has altered nearly every aspect of our personal and professional lives. As the business world reels from the disruptions generated by COVID-19, it has become more clear than ever that AI is no longer a “nice-to-have” technology for the modern enterprise.

Businesses are facing all kinds of challenges, from supply chain interruptions to rapid regulatory change to unexpected financial hiccups. Customer service is no exception, as many organizations find themselves overwhelmed by customer queries with no ability to scale up quickly to meet the demand. Mandatory shutdowns across the globe are forcing customer service departments to operate with only a fraction of their staff, while balancing the pressure to minimize costs in response to broad market turmoil. Rightly, most businesses weren’t prepared to suddenly scale their customer service operations, while also managing costs and providing both agents and customers with the technology needed for successful digital interactions.

Under these pressures, companies need to transform their disjointed customer experiences into seamless, efficient digital ones. There is a clear opportunity to apply AI to quickly answer common questions and automate simple customer service inquiries to help customers resolve their issues quickly without overwhelming contact centers, so human agents can focus on the most important inquiries. We’ve recently enhanced Watson Assistant with new capabilities and integrations designed to help support organizations as they rethink their customer service experience.

Teaches Itself to Improve: One of the biggest hurdles that enterprises face is the time it takes to manually “teach” their virtual assistants to provide better answers. In the same way that human agents need training to improve their responses, so too do virtual assistants. In the past, this task would require developers to manually look up the most common canned responses to a given query, then tell the assistant to use the most well-received responses in the future.

Currently in closed beta, the autolearning capability teaches your assistant to improve, surfacing the most relevant responses to customer queries, getting better with every single customer interaction.

Autolearning represents the latest in an evolutionary lineage for virtual assistants. First, AI was applied to formerly rules-based chatbots. Then, chatbots became even smarter by connecting to various systems across the enterprise and beyond, giving them access to vast libraries of information from which to draw responses. Now, with autolearning, these virtual assistants can improve their capabilities with limited human supervision, eliminating points of friction in the customer experience. And because your customers are having the best possible experience, they are less likely to transfer to a human agent. Agents can then be reserved for high-touch or complex customer interactions, dramatically reducing customer service costs and keeping agents engaged with more meaningful work.

Autolearning addresses disambiguation, a complex linguistic problem that presents a challenge to virtual assistants attempting to understand the intention behind a query that is not immediately clear or apparent. Now, your virtual assistant can evolve as your customers’ queries do. In a world where businesses are rapidly expanding their virtual assistants’ responsibilities, adding more entities and intents all the time, autolearning ensures that agents can keep pace with all the data being thrown its way.

No Dead Ends: We’ve all interacted with a basic chat bot or phone-based IVR where we simply cannot get the answers we need, and we either get stuck in a loop or kicked off the call. With Watson Assistant, customers never have to experience this frustration. Its Live Agent Handoff feature now integrates seamlessly with Zendesk, Salesforce and Intercom, so that even the most complex customer queries can be handled by a human agent during the same customer interaction. Watson Assistant is fully interoperable with internal knowledge bases to dynamic external content sources to your call centers and help desks, and it lives across voice and digital platforms.

Plug and Play Solution: Implementing Watson Assistant does not require a data scientist, or even a software engineer, and can be deployed in a manner of minutes. Watson Assistant can be implemented and managed by non-developers with a simple and intuitive user interface that suggests best practices and provides styling options to ensure optimal brand representation.

Customize with Unique Brand Voice: In 2020, everyone in the developed world deals with virtual assistants on a regular basis. But can they tell one from the other? We believe that the voice assistant space is going to become a heated battleground for brand identity, with organizations developing highly customized voice assistants. Offering customers a memorable experience with a character or individual they’ll remember will be a real differentiator. Watson Assistant features this ability today for voice and digital assistants, and it can be deployed with minimal training data.

The complete virtual assistant will address most customer problems without hitting dead ends, regardless of the complexity of the query. It will play well with other tech platforms. It will teach itself to improve, at scale and at reduced cost. It will do all of these things with a customizable flair, and users will be able to deploy and manage them with surprisingly little technical knowledge required. You can meet this complete virtual assistant today.

Plus, if your organization has a COVID-19 use case, IBM is offering Watson Assistant via the IBM public cloud at no charge for 90 days with no commitment required after the free period ends.