This audio was created using Microsoft Azure Speech Services
A customer-centric approach powered by AI to take customer support to another level
We often emphasize the importance of customer centricity, but it’s crucial to acknowledge a reality that many in business face: striving for perfect control is complex due to the diverse and ever-changing needs of customers. However, mastering this complexity is what truly sets you apart. This involves developing the skills, strategies, and systems necessary to understand and respond to these varying demands, consistently delivering a high-quality customer experience.
Customer support, as an integral part of the customer-centric approach, is an issue I hold dear. Why? Well, because support and care teams are often the primary point of contact between a business and its customers. They play a crucial role in understanding and addressing customer needs, which is why it’s essential to provide them with the best technology, tools, and resources to perform their roles effectively.
We prioritize the ultimate strategy of enabling self-service for simple queries, ensuring that customers can swiftly find answers on their own. However, should self-service fail to fully meet our customer needs, our live agents are readily available to provide personalized assistance, ensuring a seamless support experience. And with the integration of artificial intelligence, we are enhancing both self-service and live agent interactions to optimize customer satisfaction.
The Customer Care Knowledge Bot
When ChatGPT was first launched, I was very impressed by its ability to engage in natural human-like conversations. Unlike earlier AI models, it understood the context, maintained coherent dialogue over multiple exchanges and provided relevant, nuanced responses. The concept of having a personal AI assistant that could help with day-to-day tasks truly resonated with me. (Though I long for the day an AI assistant can handle my Starbucks coffee runs and possibly help with my household chores).
Inspired by this technology, we built the Customer Care Knowledge Bot, a secure enterprise chatbot powered by GPT technology via Azure Open AI. This is our very first GenAI bot integrated into the bFO experience to support our care agents. So far it is available to 1,2k agents and in two languages. The goal was to drive operational efficiency by providing work assistants to our agents during technical case resolutions and then for commercial as well.
Prior to this technology, agents would spend considerable time researching multiple sources of information such as FAQ documents, product data sheets, diagrams, down to the previous case history to reach a conclusion. Now, the bot research and compile all this information and goes a step further to propose what the agent could provide as an answer to the customer. Saving time, energy and enhancing speed and quality of the answer.
This technology has allowed agents to resolve cases even the ones they would have needed to escalate in the past due to knowledge gaps. They also have time to upskill and expand their expertise, which would otherwise be impossible to achieve.
Promoting High-Quality Service built on trust
At Schneider Electric, our operations are built around trust and the discussion of AI is no exception. We believe it is our responsibility to ensure the technology we build is trustworthy. Our ethical code of conduct, expressed in our Trust Charter, guides us in maintaining trust in all our relationships in a meaningful, inclusive, and positive way.
Therefore, to ensure the information provided during technical queries is accurate, our agents are required to always verify the accuracy of the responses by referring to the documents the bot provides as references. This is crucial to our promise to strive to provide high quality services, to delight our customers and protect them and their assets.
We also encourage our agents to share their genuine thoughts and experiences with the bot. Their honest feedback helps improve processes and the services we provide to our external ecosystem. This feedback loop is also essential for refining the bot’s performance and avoid hallucinations.
Best practices and key considerations
The importance of effective prompting and data quality
Crafting clear and precise prompts is crucial for obtaining accurate responses and agents need to be trained to do so. Things such as spelling errors in prompts can result in hallucinations.
Feeding the bot with sufficient information is key. A brief FAQ document is not enough to deliver satisfactory responses. Incomplete or poorly structured data in terms of format and presentation can significantly impact the AI’s ability to process and respond accurately.
2. The human element and change management
Successfully integrating GenAI technology goes beyond just the tool. Change management is essential to ensure that the agents are comfortable and proficient with the capability.
Foundational training on AI including its capabilities and limitations cannot be overstated. This will help set realistic expectations and maximize on its capabilities. You don’t use a camera to experience the moment, only to capture it. Use AI to perform tasks its designed for!
We have a long uncharted road ahead of us. The ultimate goal of this technology is its ability to handle complex troubleshooting questions, where extensive research and log analysis are required. If the bot can achieve this in seconds, its true value is realized, allowing humans to focus on building meaningful connections with customers and adding value.
Add a comment