Generative AI, which facilitates content creation through machine learning, coupled with voice biometrics (VB), which identifies and authenticates individuals by their unique voice characteristics, is transforming the contact centre. Capabilities are evolving rapidly, reshaping advanced knowledge work and leading the contact centre into a bright new future we could not imagine a few years ago.
But although these innovative tools open new opportunities for enhanced customer support and improved workforce satisfaction, they also introduce data privacy concerns, new potential for misuse and fresh uncertainty. Organisations that responsibly adopt these technologies stand to gain significant advantages in terms of customer satisfaction and the bottom line. The next two years will be pivotal in defining leaders in this space, and organisations that quickly recognise and effectively incorporate conversational AI into their operations stand to win big.
Generative AI solutions are already having a major measurable impact on modern contact centre operations which should not be ignored. Generative AI at Work (April 2023) showcases the remarkable impact of GPT-driven AI on customer care. The study reveals a 14 per cent productivity boost, reduced average handling time (AHT), increased issue resolution, lower agent turnover and a 25 per cent drop in customer escalations in one of the earliest customer-facing contact centre implementations. It is a game-changer for businesses seeking enhanced customer experience and efficiency.
Improvements in customer care, reductions in AHT and better agent retention all drive improvements in an organisation’s bottom line. In Customer Care: the Future Talent Factory (2020), the authors estimate that it can cost an organisation $10,000 to $20,000 to replace each individual agent. This clearly illustrates the value of increased employee retention and the potential for generative AI to improve the workplace.
Meanwhile, voice biometrics provides added security to customer verification processes while significantly reducing authentication time and enabling agents to help customers quickly resolve their actual issues rather than answering endless security questions. This significantly improves both the employee experience and customer satisfaction.
One international insurance provider managed to reduce the number of questions asked by 75 per cent aftervoice biometrics was deployed. The solution, which also included natural language understanding (NLU) capabilities, further helped to automate intent classification of customer calls, and provide the customer insurance number to the agent, enabling agents to address the customer request quickly and shift their focus to value-added activities such as sales.
While the focus here is primarily on the contact centre itself, the potential benefits of these conversational AI tools are not limited to this domain. In The Economic Potential of Generative AI (2023), the authors speculate that new generative AI-driven solutions in all forms may impact as many as 850 occupations and 63 specific use cases, while generating $2.6 to $4.4 trillion in new value annually, with a particular impact on banking, insurance and e-commerce.
Migros Bank in Switzerland was able to usevoice biometrics and virtual assistants to successfully authenticate its customers, reducing AHT by 20 per cent and providing agents with a visual indicator to continuously verify the caller. Thanks to this VB implementation, the system has never been compromised by fraudsters in more than five years. This hybrid implementation for both agent and self-service has generated an excellent return on investment. “Spitch’s voice biometrics system was integrated into the bank’s customer centre infrastructure seamlessly,” says Antonio Zullino, Head of Customer Care at Migros. “Identity verification by voice biometrics meets all the regulatory and legal standards, including active opt-in, accepted by the majority of our customers. This solution really helps improve customer experience while reducing call handling time”.
These solutions are a value proposition – and opportunity – that no forward-looking organisation should ignore.
Familiar use cases such as agent assist, quality management, quality assurance, FAQs,voice biometrics,speech analytics and other similar contact centre applications provide a perfect medium for hybrid solutions. Agents are already trained to manage the existing human-like challenges that these models present, while they are also well positioned to benefit from the greatest strengths of the same technologies. Recent research shows that ChatGPT, Bard and other generative AI solutions provide great value to new hires, helping to train, assist and reduce stress. Additional benefits include streamlining onboarding, increasing customer satisfaction scores and reducing AHT and employee turnover.
A leading telco in Switzerland uses speech analytics to discover problematic patterns in customer conversations. The solution provider works closely with customers to optimise the platform and improve resolution times. Issues and opportunities that are identified are then translated into agent coaching and training content to improve the customer and agent experience and boost sales.
The key to the current generation of successful conversational AI products is a customer-facing user interface/user experience (UI/UX) and a comprehensive data and analytics pipeline. From a customer perspective, this sentiment should be reflected in a judicious selection process that focuses on robust, use-case-oriented solutions backed by experienced providers. New SaaS offerings which still rely on contact centre IT to build actual solutions should be avoided. Conversely, it is important to carefully assess large-scale comprehensive providers on their commitment to delivering specialised solutions aligned to specific business needs.
Effective solution providers will focus on comprehensive use-case development and consultancy aimed at understanding specific customer needs and how to address them. Critical success and competitive sustainability with conversational AI technologies will depend on UX application design and smart integration with back-office systems. Great providers will showcase an ability to quickly iterate on product designs in response to evolving technology and client feedback.
Generative AI solutions are already starting to change the way we work, but they are also introducing new challenges. Paramount among these is the tendency of these models to sometimes confidently hallucinate answers, references and citations. In the case ofvoice biometrics there is the similar spectre of constantly evolving spoofing attacks. The fact that typical content quality from solutions such as ChatGPT and Bard, as well as recent open source software (OSS) solutions, is often both eloquent and convincing makes it all the more important that users are aware of the sometimes unreliable nature of responses that these large language models (LLMs) can return. This requires users to be both well informed and vigilant about their usage, and by the same token makes it incumbent on providers to be cautious and responsible when integrating LLMs into their solutions.
Another modern problem of generative AI is model bias. According to Princeton researcher Arvand Narayanan and his team, detailed in Semantics Derived Automatically from Language Corpora Contain Human-Like Biases (2017), this is largely a reflection of historical biases that permeate existing training data. It’s essential that providers are aware of this issue and work to ensure that their LLM-driven solutions are suitably aligned with human preferences.
Similarly, on thevoice biometrics front, it is important that providers remain constantly vigilant against emerging threat patterns.
The rapid pace of current conversational AI advancement will continue until providers, developers and corporate clients reach a collective understanding of best practices, successful use cases and the general strengths and limitations.
Businesses should expect to see significant boosts in agent productivity and retention, along with decreased AHT and better customer service, but they must also navigate technical hurdles and ethical concerns. Using conversational AI to enhance functionality rather than replace human roles is crucial. This approach mitigates issues of AI reliability and places emphasis on supporting human tasks. Business leaders operating in privacy-first sectors such as insurance and banking should prioritise on-site AI solutions.
Contact centres that implement generative AI and complementary solutions such asvoice biometrics will gain from advanced tools, leading to quicker resolutions, enhanced security and greater customer autonomy. Products centred onspeech analytics (SA) and quality management (QM) inherently involve human participation, making them ideal for addressing current AI challenges. By focusing on complementing, not replacing, human efforts, organisations can foster workplace improvements and drive true innovation. By selecting providers with deep experience in the field they can ensure that innovation leads to further success.