One New Zealand is using generative artificial intelligence to help contact centre agents understand why customers call and how to support them.
Just three months of going live, the new AWS-powered solution had helped the telco achieve a 10 per cent increase in customers who report dealing with a knowledgeable and friendly representative and a similar increase in customer trust.
One NZ developed the new generative AI solution using AWS AI services, including Amazon SageMaker, a fully managed service to build, train, and deploy machine learning models, and AWS Lambda, a serverless compute service.
The system analyses customer questions and issues generated from thousands of calls a day, and provides call agents recommendations on how to respond to incoming calls, enabling instant data-driven insights.
Agents can, for instance, inform callers about mobile black spots based on past calls.
The service integrates with cloud contact centre Amazon Connect which is used by over 1500 One NZ staff and partners across voice and chat channels. The platform is used tro answer over 330,000 customer requests each month.
“Improving customer service is a top priority for One NZ, and this requires a better understanding of our customers’ issues and concerns, and taking quick action to resolve them,” said One NZ SME and consumer director Chris Fletcher.
“Working with AWS, we’re able to use generative AI to significantly improve our speed of response, accuracy of advice, and service performance for our customers."
In building the solution, One NZ analysed thousands of previous call transcripts and categorised them by products and outcomes to assist in developing a machine learning model.
Previously, call centre agents completed record-keeping of calls and outcomes, juggling note-taking with often lengthy and complex calls with customers.
Now, the solution records, transcribes, and processes every call made to a One NZ contact centre within 10 seconds of the call ending.
Each call is accompanied by a synopsis of the product or service discussed, mood and sentiment, contact history, and the action taken by the agent.
The system also picks up language sentiment, such as sarcasm, and can process basic Te Reo Māori and a variety of accents.
Different teams across One NZ can use the large volumes of data produced to adapt products and services such as residential broadband and mobile plans to enhance customer service.
One NZ can also use it to identify emerging themes in customer issues, reduce friction for future customers, and to be proactive in decision-making.
Throughout the development, One NZ maintained a focus on ethics and data privacy, removing all personal information including names and addresses of customers with all information processed by One NZ.
The telco is now working on adding additional functionality including an interface for agents to search and access examples of customer issues in real-time on a customer call, ease compliance checks, or proactively identify potential network issues before they unfold.
“We’re going to continue evolving and developing the process, keeping customer outcomes front of mind, as this is just the beginning of how we can use data and analytics in this way,” said Fletcher.
One NZ plans to make the solution available through Amazon Bedrock, a fully managed service that makes foundation models from leading companies available through a single application programming interface.
“One NZ is on a path to becoming an AI-driven telco with customer experiences, products, and operations all driven by AI,” said Tiffany Bloomquist, country manager for commercial sector at AWS New Zealand.
“There are some compelling use cases emerging from generative AI and we continue to see great potential for it to disrupt and transform industries when delivered at scale."
Earlier this month, One NZ appointed Summer Collins to lead its AI and data development.
In its 2023 financial year, OneNZ wrote off $64.1 million of work in progress on an earlier transformation programme dubbed "Digital Accelerator".
The business support systems modernisation programme proved to be challenging and One NZ "pivoted" back to work with its existing vendors.