Artificial intelligence has left the research laboratories and entered real-world applications across many industries, delivering significant value. A good example is the enterprise virtual assistance (VA) space, where I recently created an Ovum Decision Matrix (ODM) to compare vendor products side by side.
In recent years, the use of AI in what we now call intelligent virtual assistants (IVAs) has been transformative. Through digital transformation, more business is moving online to the cloud and off the bricks-and-mortar high street, and these businesses have had to create call centres to accommodate new demand and a changed business model. The expense of manned calls has created a big demand for IVAs. The first generation of VAs may have been a cost savings for the business but were an unpleasant experience for users—and ultimately not so good for the business either. IVAs, on the other hand, offer a high degree of customer satisfaction and have a high rate of completion without deflection to a human.
There is great art and experience in how the end-to-end journey is accomplished by IVAs taking calls. First is the decision as to what kind of calls will be handled by the IVA vs the human agent. It is possible to give IVAs the easy password-reset type of queries that may make up 20% of all calls and achieve 90% fully automated completion rates. As the business decides to let the IVA handle more complex calls, the fully automated completion rate may drop, with more calls eventually passed to humans, but this is still a high ROI for the business.
Even for the IVA just being able to correctly understand a query and deflect the call to the right human agent is valuable, yielding a better caller experience and a cost savings in less wasted time. In addition, internally used IVAs have powerful use cases, such as assisting human agents as they take customer calls, correctly understanding a query and providing expert support.
End-to-end journey processes are what separate IVAs from other conversation AI interfaces, and I saw many different approaches to solve the journey challenge. Some even used humans-in-the-loop, where a human was employed as essentially a cog in the wheel, clarifying utterances and not engaged with the whole query. All the vendors I talked with had IVAs that resembled a collection of modules that performed different tasks well, using many different types of algorithms – a kind of ‘brain’ for query handling.
For the ODM I talked with Verint (which acquired Next IT in 2017). Verint’s IVA started out in 2002 with a contract for the US DoD, emulating human conversations in internet relay chat (IRC) chatrooms to catch bad guys—the IVA was built out from these origins. We scored Verint in the leader group for their conversational AI platform. We found the platform combined advanced machine learning with natural language semantic models to offer the advantages of both approaches. The solution came with a wide range of supporting tools, such as Verint Knowledge Management suite and a voice of the customer (VOC) solution that gathers and mines enterprise customer feedback.
IVA technology continues to evolve and improve, and I expect to see more use of IVAs in every sphere of activity. In my research I saw an IVA used for the in-house portal of a global enterprise. I’ve had experience searching for an item on a portal, SuccessFactors, but I spelled it “Success Factors” and received a frustrating null result. I think an IVA would have answered my query, and I’m glad to see Verint has a customer Alight with this very use case, so I look forward to seeing more IVAs in use.
Michael Azoff, Distinguished Analyst, Ovum
We hope that this analysis will help you make informed and imaginative business decisions. If you have further requirements, Ovum’s consulting team may be able to help you. For more information about Ovum’s consulting capabilities, please contact us directly at firstname.lastname@example.org.