Dick Rosinski, Executive Director of Professional Services, Nortel Networks

NewsBlast
Tell us about your recent announcement concerning Media Processing Server (MPS) 500.

Dick Rosinski Nortel Networks announced the release of its new cost-effective, self-service platform, the Media Processing Server (MPS) 500. The MPS 500 extends our MPS portfolio by providing a carrier-class solution for Integrated Voice Response (IVR) applications to medium sized contact centers and small service providers. As automation plays an increasingly critical role in our customers' businesses, there is an increasing need for carrier-grade reliability at enterprise-level size and price. The MPS family is designed to meet that need. The MPS family of MPS100, MPS500, and MPS1000 all use the same hardware and software components and offer scalable, flexible solutions to growing businesses for the first time from any vendor. The MPS500 and MPS1000 offer a very rich feature set for speech, including speech recognition, verification, and caller categorization from Nuance and from SpeechWorks, with TTS from Fonix, ScanSoft, Nuance, SpeechWorks and Rhetorical. Our administration tools allow remote update of prompts or of recognition grammars on large, distributed systems. Our application and system architecture is independent of the speech technology vendor. The MPS 500 provides IVR customers with a simple migration path to more advanced capabilities like Speech Technology and Voice over IP (VoIP). We are currently accepting orders for the MPS 500 and are planning to begin shipping the product in September 2003.

NB
What role does Nortel play in deploying speech solutions? Who are some of your partners and how did you choose them to be partners?

DR Nortel Networks offers a full set of Advanced Speech Processing solutions built on solid technology and experience. We deliver complete turnkey solutions to our customers. More specifically, our portfolio includes an Interactive Voice Response platform (IVR) with host and telephony interfaces and the OSCAR (Open Signal Computing and Analysis Resource) speech-processing platform. Also included in our portfolio are speech algorithms, speech development tools, complete applications, project management, training, installation and ongoing maintenance/enhancement services. We have created and deployed several dozens of customized advanced speech recognition applications for brokerage, transportation, banking, insurance, public utility, government and telecommunications service provider markets. In addition, we have an application portfolio of what I call "replicated solutions," which consists of the largest number of standardized applications in the industry. These are solutions for which the design, coding and testing have been completed; all that's needed are specific prompts or host interfaces to deploy an application in a customer's environment. Of course, we can customize these standard applications to reflect our customers' wishes. Our Speech Technology Portfolio not only provides the customer with significant benefits, but our solutions also help them to improve employee productivity, lower operational costs and increase revenue. Most importantly, Nortel Networks customers report significant reductions in transaction costs and improved customer satisfaction as a result of deploying a Nortel Networks speech recognition solution. Nortel Networks and our partners are committed to leveraging our customers' investment in our Advanced Speech solutions whether it is with our packaged solutions, customized offerings, superior technology or customer service excellence.

NB
As partners with both SpeechWorks and Nuance, how do you see the ScanSoft and SpeechWorks merger affecting this market? What does their transaction mean to customers?

DR Consolidation in technology markets is not unusual, and we are seeing it now in the speech industry. From the beginning, Nortel Networks anticipated changes in technology and we have implemented architectures to shield our customers from any adverse effects. We have always had a common API that is used to integrate all of the various speech engines into our platform. We have our own tools for development, performance analysis & tuning, and testing. Our application development processes are the same regardless of engine. What this means is that there is very little in a Nortel Networks offer that is specific to a technology vendor. We have even moved a customer application from one recognition engine to another; the effort took about four days for the port and a week to regression test. Another customer is using one recognition engine at one place in an application and a different engine at other points in the same application. These aspects of our product allow us to assure customers that we will offer leading edge technology across inevitable change.

NB
What are your thoughts concerning the various standards that could impact speech technology? Does standardization really help customers and do they demand standards to be in place before they deploy technology solutions?

DR I strongly support positive current standards efforts. The timing and pace of the standards efforts have been extremely useful for the industry. Standardization is a great help both to customers and to vendors. For one thing, they open the marketplace. If I want to move a customer's application from a competitor's platform to mine, software standards could make that a relatively inexpensive port, rather than a complete redevelopment. Standards won't eliminate the idea of a forklift upgrade, but perhaps they'll require a smaller forklift. Since standards are constantly evolving, customers don't, indeed they can't, wait for a standard to be in place before deploying. Speech results in improved customer satisfaction, better service, and reduced costs. It doesn't make business sense for customers to forego these advantages to wait for a standard. What's needed is a way to migrate with the standard as it develops. For example, let's discuss VXML. We have thousands of customers with applications written in PeriPro, our GUI development language, and we support the VXML 2.0 standard. Our PeriPro and our VXML environments are completely compatible. A VXML application can call a PeriPro application or module, and vice versa. What this means is that there is no need to supplant the VXML standard with proprietary extensions. A customer can have a VXML application that handles call processing and call control, or legacy hosts by using their existing PeriPro modules. This allows us to support emerging standards while protecting our customers' investment. A customer can do its own migration planning secure in the knowledge that our systems provide backward and forward compatibility with the standard.

NB
We hear a lot about how customers are not served when organizations automate customer service. What do you think is the right balance between great customer service and automation?

DR Well, I don't think that great customer service and automation are polar opposites. Often, great customer service comes from automation. As a consumer, I often want to transact some business; I don't want to talk to an agent, I want to pay a bill, get some information, make a reservation, order a product, etc. If I can do that quickly and easily, whenever I want to without having to wait in an agent queue, then that's great service. A company provides great service if it uses its agents to handle issues that require judgment and human intervention, rather than for reading a script in common transactions. I think this is true throughout the history of technological innovation. Operators setting up a call vs. automated call setup; steno pools vs. text processing; talking to a librarian vs. doing a Web search. Uniformly, automated services are better, faster and cheaper. This level of automation is just appearing in the call center and telco service sectors, but the principle will hold here as well. To provide great service, automate the frequent or commonplace and use people for their intelligence and helpfulness. The integration of our self service (MPS) products with our call center (Symposium) products makes it easy for customers to create the right mix of automation and agent care for their business circumstances.

NB
Compare speech solutions vs. other forms of customer service automation such as Web based, touch-tone and live agent support. What are your customers telling you about speech vs. other modes of customer interaction?

DR What we are clearly hearing from our customers is that all of these modes are essential to them. Although they offer similar capabilities, the various technologies are not simply interchangeable. We are seeing patterns of customer interaction emerging in which an end customer might shop for a product on the Web, order using speech, check on store location using Touch-Tone and talk to an agent to resolve an issue. No one would imagine forcing customers to deal with only one modality. That's part of the reason we refer to our product family as Media Processing Servers rather than IVR or Voice Response Systems. Our applications can live on a web server, or on the MPS; they can use speech or touch-tone; they can automate an entire transaction, or act as a front end to an agent center using CTI or CRM.

NB
Many in the speech industry certainly expected to see more speech applications by now. What are the hurdles that need crossing to increase the adoption rate of speech?

DR We have applications deployed on hundreds of thousands of ports taking millions of calls. Industry growth rates are no longer exponential; they're more like the growth rates of established technologies. But there are three hurdles that face the industry. First, in Geoffrey Moore's terms, we have crossed the chasm from early adopters to the mass market. Customers now are less impressed with the "Wow!" factor, and more concerned with business goals. Nortel Networks sales and professional services work closely with customers in pre-sales, discovery, and design phases of a project to set project goals and to measure performance. Second, we need to realize that speech technology is a critical, but a small part of deploying a successful application. Speech applications can be business mission critical. System integration, reliability, scalability and investment protection are what are needed to make this technology truly useful in a business context. Third, in some cases there has been a reaction to the exaggeration and overselling that is sometimes seen. Customers remain skeptic and rightfully so when they read and hear statements of how speech improved call completion by over 100 percent, or how speech reduced costs by almost 100 percent. The business rationale for speech is very powerful. We need to show customers that the technology works and that the application and system integration insulate them against any risks.

NB
Tell us a little about your background and how you ventured into speech?

DR I joined Bell labs in the 1980s initially working on network services that used Touch Tone signaling and compressed speech prompts; services like network teleconferencing, voice mail, Advanced 800. When speech recognition technology was ready, it was an obvious input technology for business services. AT&T deployed speech recognition in 800 services and in operator services, and in the mid 90s my department conducted very successful trials of recognition for automation of business transactions. We offered Advanced Transaction Processing Services via the InfoWorx service bureau. When AT&T decided not to pursue that business direction, I joined Periphonics to head their Advanced Speech business unit. Subsequently, Peri dominated that market, and speech is an important part of our business in Nortel.

NB
What suggestions would you provide to companies who are implementing speech solutions?

DR Just as in the old cliché about real estate, the three most important things about implementing speech solutions are experience, experience and experience. Select a reliable platform offered by someone experienced in integrating complex technologies in telecomm or call center or web environments. Focus on your core business, not on technology. Assign people to the project team who understand the business rationale for your project. Define your project as a series of business successes, rather than a big technical splash. I'm sometimes concerned when I hear about do-it-yourself efforts in the industry. People expecting to buy hardware from one vendor, software from another, and applications from a third are destined to spend a lot of time learning how to do systems integration. If you are adopting speech technology to solve a business problem, take the easy route: Go with Nortel Networks.