Astute Aims to Make Virtual Agents Seem More Real

It’s the Holy Grail of contact centers — creating a virtual agent that performs like that “real, live human being” that so many of us hope to reach when contacting a company. Unfortunately, the gap between automated agents and human ones remains a large one, with voice recognition and language processing technology lagging behind customer expectations.

Now, knowledge management software maker Astute Solutions has announced the the release of version 3.0 of its RealDialog e-service application, which claims to make a quantum leap in the realm of the virtual agent.

The software provides customer interaction through a range of channels, including websites and voice contact center systems. It can, says the company, deliver responses to customer or employee inquiries in a variety of forms, including voice response, document retrieval, and email responses to common questions.

Responsive Responses

RealDialog 3.0 claims a 99 percent accuracy rate in responding to customer inquiries. That level is measured by the software itself, Laurie Angalich, vice president of marketing at Astute, told CRM Buyer.

Customer companies see a level of about 95 percent accuracy of response within a short time after deployment, she said, followed by a jump to nearly 100 percent after a couple of months. Determiners for that accuracy include whether or not a customer asks a follow-up question and software-internal analysis of whether or not the answer is indeed responsive to the language submitted by the customer.

The difference between RealDialog’s language processing basis and that of other IVR and automated agent systems, said Angalich, is the fact that its analysis of customer inquiries is based on more advanced linguistics.

The original iteration of the software was developed by linguistics professors, she noted. It uses sophisticated analysis of phrases rather than the typical analysis of keyword patterns employed by most automated agent software.

By analyzing customer inquiries word for word, said the company, RealDialog “learns” from the question and answer pairs from every user conversation. The application can analyze thousands of unstructured questions and conversation and identify content gaps in contact center and Web repositories.

The software extends this language-processing capability to deployment, populating the knowledge base in part by “learning” from a company’s existing Web pages, documentation and enterprise content.

Thus, said Astute, non-technical business users, such as brand managers, can add, modify and tailor responses, content and dialogs for multiple touchpoints using text and HTML. Through content collaboration tools, they can delegate questions to subject matter experts.

Automating the Social Network

Because it is based on a knowledge management platform, RealDialog creates two-way communication, noted Angalich.

For example, companies often are receiving lots of feedback from customers through social networking channels such as Twitter and Facebook presences. If many customers are asking similar questions or submitting similar negative comments about a particular product, those comments can be aggregated and sent to the appropriate brand manager or customer service manager, she explained.

The person responsible for developing the response to such inquiries can then formulate a “brand-approved” response and populate contact center and Web response channels with that response.

Astute will be making another product announcement in the very near future regarding social networking and virtual agents, said Angalich.

RealDialog 3.0 currently is deployed at a number of enterprises, although Astute declined to name them. Publicly, however, Astute counts among its users Bath and Body Works, Unilever, Miracle-Gro, and Water Pik.

Pricing varies considerably by installation, Angalich noted, and the software is available both for on-site installation and on a subscription basis.