Agile Language Partners

Intelligent Agents

Complex neural networks are being linked to interactive agents. IBM’s Watson seamlessly integrates a natural language processor with an advanced neural net. It then can interlink this to AI driven automation. “Watson and other intelligent agents are remaking everything that depends on natural language processing, complex processes, and access to structured or semantic data,” from the Promise of Artificial Intelligence in Telecom. Most of this is still experimental and promotional demonstrations. But in the applications to which it has been applied, it demonstrates exceptional power. IBM is not alone as Facebook and Google are working feverishly to this end. This is far beyond process flows. This will replace workflow. This has the potential to obsolete almost all OAM&P product lines that do not adapt.

How can an ICE company leverage this? Strategically, get ahead of the transformation and drive it. Some of the big players, aka IBM and Microsoft, are already telecom partners. Many billing and customer management software companies (Amdocs, Metratech, apptium, CSG, Ericsson, and others) are experimenting with AI for improving customer service. Software-Defined networks (SDN), Network Function Virtualization (NFV), and Self-Optimizing Networks [SON] are creating interfaces to the network that AI can access.

In practice, ICE executives can implement a three-phased strategy to gradually adjust to these technologies while taking smallish steps to significant agility.

First, obsolete IVR systems need to be replaced with the current Intelligent Assistant services. These are both voice-activated and platform resident. The same old VXML interfaces could be accessed by Intelligent Assistants. But direct integration to process control, workflow, and back end systems is probably more efficient. For example, ICE companies could develop “skills” on Alexa that allow an Echo household to request a service call or order a new service.  This could be as simple as “Alexa, subscribe to HBO for one month via my Verizon account.” Or “Siri, enable international texting in Spain until next Thursday.” Of course, the transaction request automatically is recorded – non-repudiation. It could be authenticated via an SMS request to enter an authentication code.

Multilingual Intelligent Assistants have another customer service advantage.  They can recognize the key phrases which signal the language the caller is speaking. Then a simple confirmation response, in the expected language, personalizes their experience. No more forcing all customers to experience delays for language responses they will never push.

Second action, Intelligent Assistants need to be authorized for daily use by all internal employees. Of course, this should be rolled out incrementally to different specialized activity groups as the agent skills are developed. Initially customer service agents can access the same Intelligent Assistants as are being provided to the customers they service. They could even trial new skills internally before rolling them out to the customer.

That is only the threshold. Every skill set in telecom could progress to be more agile by enabling assistants to help employees with their routine tasks. Agents could work with NOC engineers: “Cortana, pull the configuration file from router AB49.  Check it against the last backup.” And as the AI learns the engineer’s typical solution approach, it might automatically pull the config files from adjacent links where a change was found in that backup comparison. Eventually, with NFV/SDN it would allow a traffic management engineer to train an assistant to prioritize traffic. Then a sales rep could be at a customer meeting, get a live customer request, and request his agent to “Prioritize video on International Widget’s link from New York to Chicago during their analyst meeting”. The productivity of every employee would skyrocket.

Third action, as new skills are learned by these internal Intelligent Agents, with proper filtering and authentication, they can be offered to the customer service agents and eventually directly to the customers Intelligent Agent Interface. More skills for the customer equals greater agility.