Capitec is looking to incorporate more generative artificial intelligence (AI) functionality into its customer service processes as a way to help agents resolve customer queries faster.
The use of AI and other technologies in the customer services or business process outsourcing arena has grown over the years. So much so, that interacting with automated chatbots to deal with certain customer queries has become commonplace in SA and many parts of the world.
Imagine a world where human customer services agents have access to their set of generative AI tools and chatbots, akin to OpenAI’s ChatGPT, that “whisper to” call centre agents as they deal with queries, offering real-time assessment, guidance and advice on how to quickly resolve an issue.
This is a future that Blessing Mgaga, divisional executive for retail client experience delivery at Capitec, says is not too far off.
“We’re not far. We’re actively working on some use cases, internally. More and more, we see the application of AI as a [kind of] co-pilot, supporting our agents and in our branch ecosystem as well,” Mgaga told Business Day during Amazon Web Services’ recent SA summit in Johannesburg.
AI systems can provide agents with real time access to relevant information, knowledge bases and suggested responses during customer interactions. This can help improve efficiency, accuracy and consistency in handling customer complaints and queries.
Once a call or interaction is done, these same systems can generate summaries of customer calls. AI can also analyse customer conversations to identify and track sentiment and emotions in real time, helping agents gauge customer satisfaction, identify potential issues and tailor their responses accordingly.
“We’re looking at those types of solutions, because we can ground the models that we build on Capitec data [and] make them safe. We already have the data repositories and sources. We already have the technologies on Amazon Bedrock. So, we can feed those into the channels that our agents work on and they can quickly chat to it like you chat to a chatbot today,” Mgaga said.
“Right now, you can chat to us through WhatsApp. Same with agents. If they’re on a call with a client, the information is already fed to our model systems, giving history of what has happened with the client and their specifics, what they’re likely dealing with and how to effectively deal with it.”
However, Chris Erasmus, country GM for SA at Amazon Web Services, said “it’s not about whether the technology and models can support [this]. The technology exists to be able to do all that. If your data is in order, then it’s absolutely capable. It’s more about the trust.
“Like they say in Spider-Man, with great power comes great responsibility. The technology is there. But even with all the capability, how much do you trust it to do all that without the human confirming and actually providing the validation.”
Generative AI models, especially those trained on large sets of data, can sometimes produce inaccurate or misleading information. This can lead to customer frustration and damage to brand reputation.
AI, though capable of mimicking human-like interactions, lacks true empathy and emotional intelligence. As such, it may struggle to handle sensitive or complex customer situations that require a human touch.
“We now have teams whose job it is to work on the data and make sure that we publish the data that our agents see and are able to us to respond to specific client needs, to keep the quality of data at the right level to manage that trust,” Mgaga said. “Again, the technology is helping us to get there faster.”
Would you like to comment on this article?
Sign up (it's quick and free) or sign in now.
Please read our Comment Policy before commenting.