Global CSPs need a language-agnostic AI operations assistant

Commercial context

CSPs often establish back-office centers in multiple countries which can cut costs by up to 10 times through tapping into skilled labor at lower wages and helping the CSP to expand globally, entering new markets and serving diverse customers more efficiently. However, shared operations centers can struggle with language barriers, and acquiring talent with the local language skills required for accurate communication with customers can be a challenge. To optimize communications, data processing and content management across multiple geographies, then, CSPs need language-agnostic operations processing capabilities.

The solution

This Catalyst is therefore developing an AI-based business operations assistant that can provide reliable, accurate and contextual solutions in the customers’ and agents’ preferred languages. The goal is to provide customers with a better communication experience: they won’t have to repeat themselves to be understood, or put up with ambiguous responses because of language and dialect differences. The assistant can also support locale-agnostic data processing and multilingual content generation.

The solution integrates multiple channels of communication such as emails, live chats, calls to contact centers, and interactions through portals and applications designed exclusively for partners and vendors. It uses generative AI (genAI) translation and classical conversational AI tools to take into account contextual information such as brand voice, tone, industry and audience, so that translations are both accurate and on-brand. The underlying large language model (LLM), which is being trained with process specific terminology to enable contextual and accurate communication, will become more accurate over time.

Developed as a TM Forum Open Digital Architecture (ODA) component, the Catalyst makes use of IG 1321 ODA Component Conformance to design a modular solution on AWS according to the principles of the ODA Canvas reference implementation. The genAI solution has been developed in accordance with IG 1335 Gen AI checklist and uses the IG 1352 ODA Telco to TechCo AI driven Enablement Framework.

The assistant will be available as a service on AWS, with modular components that will make it simple for it to be white-labeled and customized. It is designed to integrate easily with data stores and applications belonging to various regions, with their own data formats, languages and other constraints, and ensure low-latency processing of information and responses. The assistant will work in secure AWS environments with no storage of personal identifiable information, supported by guardrails to ensure responsible and bias-free responses.

Applications

The language-agnostic assistant is designed to enhance productivity, speed up problem resolution times and increase customer satisfaction. The Catalyst partners anticipate the assistant will deliver initial productivity improvements of 20-30%, an increase in NPS and CSAT scores of 5-10 points and a 5-20% reduction in resolution times.

Wider value

The Catalyst’s unified language-agnostic approach is designed to replace region-specific systems and reduce the need for specific language skills among staff. By streamlining and simplifying communications, CSPs should therefore be able to achieve greater inclusion and diversity, both in terms of their customer base and their own workforce.

“Can you imagine finding a Greek-speaking employee in India with accounting skills?” says Gjurgjica Gina Pesheva, Senior Innovation Expert Order2Cash at Deutsche Telekom. “This language-agnostic solution approach presents invaluable advantages for our company's offshoring endeavors, particularly in payment and collection processes. It addresses challenges in recruiting, training, and retaining staff in offshore locations that can work in our processes without the language barrier. With this approach, we will be in position to establish offshore business global centers serving customers in different languages, without the need to have those language skills in the offshore locations. This solution is applicable in the wider industry for any company that has centralized operations across different countries or plans near shore or offshore operations.”