From hype to impact: How CXone Mpower Gen-Insights bridges the generative AI divide

Generative artificial intelligence, or gen AI, has been heralded as a revolutionary force, poised to transform industries by driving efficiencies, enhancing creativity and improving customer experiences (CX). As organizations rush to adopt these new technologies, the promise of generative AI may starkly contrast with the reality of its implementation. While the potential is immense, many companies struggle to realize the full benefits due to various challenges.

We will explore how to bridge the gap between the promise and reality of gen AI, focusing on best practices for successful deployment; and highlight how solutions like NICE CXone Mpower Gen-Insights enables enterprises of all sizes to improve customer satisfaction while boosting efficiency and reducing costs—at unprecedented scale.

Would it surprise you that conversational AI has been around for over 50 years? It has reached its current level of sophistication only in the past decade due to advancements in machine learning and deep neural networks. In the 2010’s voice-driven personal assistants like Google Assistant, Amazon Alexa and Microsoft’s Cortana gained wide adoption. These have become fairly ubiquitous in households for their utility albeit niche focus. The rise of transformer-based models like OpenAI's GPT series revolutionized conversational AI. With the release of GPT-3 in 2020, followed by ChatGPT in late 2022, conversational AI reached new levels of fluency, understanding and application, making systems capable of generating new text that closely mimics human dialogue.

We can all likely recall the first time interacting with these technologies with great interest and high expectations, only to be disappointed with no response, a wrong answer, or unhelpful instructions. The reasons for this are many—the training data was gathered casting a wide net with public sources that may have contained incorrect or conflicting information, and the public had a limited understanding of the concept of prompt engineering. In hindsight, it’s easy to see why our simple or unclear inputs returned simple or unclear outputs. With the rise of prompt engineering roles in professional settings in the last two years, it’s becoming clear gen AI can be beneficial to CX across multiple industries when prompted correctly and using the right data.

The promise vs. reality of generative AI

Gen AI has the potential to significantly boost productivity, with McKinsey suggesting a 30%-45% increase when applying gen AI to customer care functions. Its capabilities extend beyond automation, offering the possibility of personalized, efficient and high-quality customer interactions that can drive loyalty and satisfaction.

Gen AI is not just a buzzword; it's a tool that can fundamentally reshape customer experience. According to recent Aberdeen research — more than twice as much as service delivery efficiency, or reducing operating costs- the top priority from managers to C-Suite is improving CX results and consistency.^[1] However, the reality is that many organizations face significant hurdles in realizing these objectives.

Common pitfalls

One of the most common mistakes companies can make is underestimating the nuances of applying AI to CX initiatives. Many assume that generative AI can be a one-size-fits-all solution, leading to unrealistic expectations. This "generative AI can do it all" mindset often results in disappointments when the technology doesn't deliver as anticipated.

Forrester also highlights the importance of understanding the implications of poor execution of AI to avoid strategic missteps: “A company relying on information from gen AI to make a critical business decision will stumble in the market if the information is erroneous or fabricated.” Without this knowledge, companies may face issues related to poor CX, high costs and ineffective AI applications.

By now we have all seen examples in the news like competitor products being recommended and refund policies being made up, where the adoption strategy has had a dramatic impact on CX. The possibility of hallucinations and rogue responses such as these require careful consideration and thoughtful implementation because brand loyalty and trust must be protected. Brands that do can gain a competitive edge.

What are best-in-class companies focused on? Aberdeen recent research confirms that the top-performing organizations are 74% more likely to already be using or planning to use generative AI for data insights and analytics compared to all others. Brands without a cohesive adoption strategy will find it hard to compete. ^[1]

Strategic framework for success

To avoid these common pitfalls, organizations could adopt this strategic approach:

• Selecting the right LLMs: Choosing the correct large language models (LLMs) is crucial. The effectiveness of AI largely depends on how well the chosen LLM aligns with the specific needs of the business. A tailored approach ensures that the AI delivers actionable insights rather than generic outputs.
• Guided prompt building: To maximize the benefits of generative AI, it's essential to have guided prompt building. This involves crafting precise prompts that minimize risks and yield the most relevant responses from the AI.
• Purpose-built for CX AI models for industries: Industry-specific models offer superior accuracy and relevance compared to generic LLMs by leveraging specialized training on vast, domain-specific data, enabling them to generate more tailored insights and custom data types automatically.
• Unified platforms for enhanced CX: A unified platform that integrates all customer data points is vital. Such a platform ensures that AI-generated insights are not siloed but accessible across the organization, leading to a more cohesive and effective CX strategy.

Introducing Gen-Insights

Gen-Insights is a guided prompting solution designed to harness the power of generative AI to reduce manual effort with accurate automation. This solution combines pre-built customizable prompts with guided prompt engineering, industry-specific models addressing the unique needs of different sectors, and uses best-fit LLMs optimized to the business’s needs within a unified platform to deliver actionable insights that reduce cost and enhance CX. Unlike other solutions, Gen-Insights is powered by Enlighten AI models purpose-built for CX, business-specific criteria, and creates seamless workflow automation across CXone Mpower—providing a complete, connected, and intelligent CX AI platform for use cases like quality assessment automation.

In many organizations, the process of assessing quality across customer interactions is both complex and labor intensive. Traditional quality assessments often involve massive manual effort, making them cumbersome to complete and analyze. This consumes valuable time and resources and introduces the risk of human subjectivity and bias, leading to potentially inaccurate results and oversight of critical issues.

As an example, let’s imagine a travel industry business. There are only 40-50 people for quality assurance (QA) covering 1,000-plus agents. Their available time for meaningful coaching is limited, and they can only evaluate a tiny percentage of total interactions. Even though the quality team has resource constraints they still need to ensure accuracy, objectivity, and good coaching to increase operational effectiveness. This example likely translates to poor visibility and CX for the business. If this sounds familiar you are not alone. There are many instances of companies constrained to very small sample sizes, sometimes less than 1% of interactions, that limit their ability to understand the full picture. They could benefit from automating manual evaluation tasks so their limited time could be focused on coaching to drive results. However, deploying a generative AI solution using a do-it-yourself generic LLM and prompt engineering from scratch will likely result in extensive trial and error, unnecessary costs, and may still return erroneous results because the training data is too broad. It seems they are caught between a rock and a hard place.

Generative AI, when harnessed through Gen-Insights, offers a powerful solution to these challenges. By automating the analysis of complex quality assessments from vast amounts of individual interactions, Gen-Insights significantly reduces the manual effort required in this process. The solution guides the user and simplifies building optimized prompts with minimal expertise required to evaluate customer interactions based on user-defined business-specific data and industry-specific models. This automation ensures that all relevant aspects of quality are consistently accurate and reliable without the need for extensive human effort, enabling businesses to quickly improve quality programs and realize positive business outcomes.

Realize true potential of gen AI

Bridging the divide between the promise and reality of generative AI requires a strategic approach that combines the right tools, knowledge, and practices. By following best practices and drawing on the ultimate AI hyper-platform like CXone Mpower, organizations can unlock the full potential of generative AI to enhance customer experience while reducing manual effort and cost.

Explore how NICE can transform your operational efficiency and help you realize the true potential of generative AI.

^{[1] Aberdeen Strategy & Research: State of the Contact Center (2024)}