There’s a lot of talk about using generative AI in marketing and sales, but what about customer experience? Yes, chatbot experiences exist, but that’s just one way to leverage generative AI. We spoke with two experienced technology leaders to better understand the opportunities in areas such as customer support, contact centers, and digital customer success.
Mladen Milanovic is Vice President of Automation at Presidio, a global digital services and solutions provider. Milanovic has extensive experience in contact centers and customer experience. When Presidio decided to build out its automation practice, it determined that its greatest synergies and impact would be in its contact center. Milanovic says automation has had a significant impact on contact center efficiency, including increased agent satisfaction and productivity. Most importantly, customer satisfaction has improved.
This contact center was an early adopter of AI, including natural language processing (NLP) and machine learning (ML). Milanovic argues that AI technology has become almost a commodity in the industry. This technology is proven, has many best practices and strong his ROI. So it’s only natural to start thinking about:
One of Milanovic’s areas of interest is agent-assisted technology. For example, sentiment analysis systems can provide feedback to agents on how to handle situations. While this is still NLP, Milanovic believes generative AI and Large Language Models (LLM) will accelerate adoption of this kind of technology and provide even better results.
According to Milanovic, contact centers often provide the only interaction brands have with their customers. It means that dialogue matters and that information is of high quality and accurate. This is what Milanovic believes will lead to the adoption of generative AI in internal systems, such as agent-assisted software, rather than virtual agents that interact directly with customers, at least until the challenges of inaccuracy and hallucinations are resolved. The reason is.
Milanovic argues that there is no way to effectively control, monitor and manage AI agents in real time. There are ways to tweak and tweak it, but the tools are technical and developer oriented.
This means contact centers need an ecosystem or framework that allows business users to easily track performance, measure and fine-tune virtual agent performance in real time based on historical data. It may respond and intervene more quickly to enhance or modify, adjust, and retrain agents. Today, this is a highly technical job and requires a very scarce workforce from an application development perspective.
He added that new low-code/no-code tools are emerging, but they are still new and unproven, so contact center executives are unlikely to put anything on the front end that they cannot control or measure. However, on the backend, we plan to experiment with tooling for agents.
Generative AI needs firm control over enterprise data
Michael Ringman, CIO of TELUS International, suggests there are other ways generative AI can be leveraged for customer experience, such as chatbots and language translation. TELUS International is a spin-off of his TELUS focused on serving customers across multiple lines of business, including many of his experience capabilities.
Ringman says many customers are asking how they can take advantage of new technologies such as generative AI to move forward. According to Ringman, when the customer base is large, the conversation is always interesting, and the answer depends on where the customer is in the business’s portfolio.
I think it’s the data that really matters, right? Accessing data, understanding the knowledge of data, and making the most of it is the very essence of driving Gen AI and leveraging it for better functioning across your organization. You can’t just put Chat GPT as a new chatbot on your internal or external support site and expect all call volume to go away. That doesn’t work.
Enterprises should better manage their data instead of focusing on building LLMs. Ringman argues that there is an untapped well where the voice of customer data lies dormant. His company uses tools like Google CCI, Bard and other extensions to speech-to-text to collect and mine data.
He added that many customer support teams and bots (choose your own adventure type) use outdated knowledge bases. LLM and generative AI can have more free-handed conversations, but hallucinations are more likely if the dataset is not properly tuned.
Usually what we start with these customers is let us help them get some of this unused data. Because the data we have today, especially when it comes to customer experience, is usually close to average handle times and typical traditional call center type statistics. And it doesn’t really get to the heart of the matter of meeting customers where they want to meet and understanding what they’re looking at.
The rise of the AI generation isn’t necessarily new, he points out.
This is the same challenge call centers have had for a long time. What’s interesting is that gen AI brings even more tools to your toolbox to tackle more complex problems and actually start providing personalized service to your customers. Get hands on and start giving a more human feel when robots interact with these bots, providing a truly differentiated experience. ”
humans are always in a loop
Ringman and Milanovic agree that customer experience always requires human involvement.
When Presidio advises clients on using generative AI, the company recommends providing additional training to its agents. Milanovic says it’s important for agents to feel comfortable disagreeing with her AI’s recommendations rather than relying entirely on them. There are also examples of bad results from full reliance on AI.
Ringman points out that IVR is seen as the end of the “traditional” contact center, after which chatbots will take over. He suggests that while the best of intentions exist, people are still needed.
This is because we now have these large language models, making interactions with bots more realistic and more authentic. Just because we have access to and understanding more datasets doesn’t always solve all those challenges. ”
Commenting on the need for responsible AI, Milanovic said:
Everyone is talking about it now, but there is no overall framework to guide an organization on how a tool that is in a very specific stage of development should be used. I think we need to really create and help our clients create those frameworks and have checks and balances around that.
On the other hand, Mr. Ringman argues that:
It’s all about understanding how we build these datasets, reduce the bias across datasets, and reduce all the corruption that can potentially lead to those datasets. Again, it can be said that the human participatory aspect of this will continue to exist. Jobs will change, but they’ll still want to be up-to-date about it.
Also, Milanovic said, as long as human labor becomes cheaper (most contact centers employ agents abroad where labor costs are cheaper) and the cost of leveraging LLM and generative AI is higher, adoption will be lower. claims to be deaf.
As technology becomes more accessible and commoditized, will adoption increase? But right now, AI is a hot topic in the boardroom, along with security, and everyone is just reading a book. [article] title. In general, no one reads this article and understands the limitations.
my view
Contact centers remain a key component of customer service and customer experience, with a continued focus on improving how agents support customers. Milanovic and Ringman agreed that the first part of using generative AI would be to support agents, but as trust grew, it would move to the front-end customer experience as well. Chatbots, or virtual agents, are getting better (Conversica shows how this happens).
But I agree with Ringman that data should come first. We never seem to stop talking about eliminating content silos within an organization. And until that happens, organizations will be unable to understand their customers better than they ever have, leveraging AI, let alone any technology, to create the best possible experience.
Next, we’ll look at some additional ways generative AI can improve the customer experience, especially by supporting back-end workflows and processes.
