When AI powers your customer service with a copilot.

This is the position today. In the future, this may all change as developments are ongoing. Meta is now in the game with their Llama system and ChatGPT 5 is rumored to be coming out this summer. New developments are expected to include the Sora text-to-video system and AI agents with contextual understanding. This latter development could be important as it may usher in the normalization of digital AI assistants. We all know how long it can take to book a flight online because so many variables are involved, and small differences - like changing the departure date by one day - can dramatically change the price. If you can ask your digital assistant to do all the research and booking, then this could lead to customer service designers needing to think in more detail about AI-to-AI interfaces where the customer is allowing their AI assistant to operate on their behalf. But the use cases that are still talked about most often in the customer experience environment are:

Improved chatbots.

Using language models to train chatbots on large amounts of product data - such as user manuals – so the bot should be capable of answering any question where the answer will be located in the documentation. Instead of needing to plan the '30 most frequent questions', the bot can create an entirely new answer based on the training data.

Contact center automation.

Removing manual tasks and processes from the contact center so the agents are supported by digital assistants. They don’t need to write notes about calls or decide how to categorize them and problematic calls are automatically highlighted to a team leader. All this can help agents to focus more on the customer, rather than the administration work they usually need to handle. I’ve read about these cases repeatedly over the past year, but there is another approach that can improve contact center operations beyond removing mundane and repetitive tasks. In many cases, the customer will still want to speak to a human. Maybe this will also change eventually, but right now, we are in a transitional phase where humans often expect help from another human when they call for help. It may also just be easier for complex problems – no matter how complex the question is, the human will understand. We don’t need to train the customer in how to write prompts that will be easily understood by our AI system. So, if the focus is on using AI as a co-pilot to help the agent, then there are several very strategic and interesting areas where we can go far beyond just writing up a call transcript.

Knowledge.

The agent can’t possibly know about every single product the company sells. Imagine an electrical goods company selling fridges. The agents can’t be expected to know the physical dimensions of every single fridge the company sells, but AI can. Load up every single detail of the products, policies, and systems and your AI co-pilot can be there advising the agent on difficult questions with facts that come straight from the manual. For example, imagine your fridge needs a replacement. The AI assistant can assist the agent by suggesting all possible models that fit the available space. It can look up the exact dimensions of the old fridge and then filter through the company's inventory to find a perfect match. Additionally, it can consider other factors such as energy efficiency, customer reviews, and current promotions.    This ensures that the agent can provide a tailored recommendation quickly and efficiently, enhancing the customer experience by reducing the time and effort needed to find a suitable replacement. Another example is when my colleague wanted to surprise her husband with a new bike rack. She initially checked a manufacturer's website but couldn't find a suitable model for their car. The customer service agent could only confirm that they didn't have a bike rack for that car model. Later, she discovered that no rear bike rack exists for her type of electric car due to design constraints. If the agent had a digital co-pilot, they would have instantly provided this information and suggested alternatives, such as a roof-mounted or hitch-mounted rack. This efficient response would have saved her time and frustration, allowed her to choose a suitable bike rack quickly, and the manufacturer would get a happy customer.

Languages.

Your customer support operation can immediately manage a multitude of languages if they are supported by AI. This can be really useful for entering new markets or managing markets that use languages that are very hard to support. The AI assistant can also support agents with regional differences. Imagine an airline facing a sudden surge in calls due to a snowstorm disrupting flights between Oslo and New York. Many customers calling for assistance are Norwegian speakers, potentially overwhelming the airline's limited Norwegian-speaking staff. With AI-powered translation, an English-speaking agent can seamlessly handle these calls. The AI instantly translates both sides of the conversation, allowing the agent to understand and respond to the customer's needs effectively. This not only reduces wait times and frustration for Norwegian-speaking customers but also prevents the support system from becoming overloaded, ensuring timely assistance for everyone. In this way, AI allows the airline to maintain high-quality customer service even in unpredictable, high-pressure situations.

Flexibility.

Your agent has all the product and policy info and can check anything with their assistant, but they also have the flexibility of a human. If the customer has a fridge that’s a week out of warranty and it needs a new motor, then the human can make a judgment call on the caller – are they calling regularly with problems? Do they seem genuine? What’s the value of helping this customer immediately rather than insisting they buy an extended warranty? The idea of a digital co-pilot is attractive for many additional reasons that go beyond improved support to the customer. It means that agents can get up to speed faster, so onboarding is easier, and training is also easier because the AI can advise where each individual agent needs help. It’s like creating the opportunity to supercharge agents. They are still offering human-to-human contact and flexibility, but they know everything about every product in every region and can instantly use the customer's local language. Contact center automation with AI has largely focused on reducing manual processes, reducing cost, and enhancing agents' focus on the customer. The digital co-pilot approach takes this much further and suggests that humans will remain at the center of the customer experience for some time to come. Still, agents themselves can use AI to improve the service they offer to customers. It also supports the agents so they feel better about their own role too – the work is more satisfying. AI will get better and better, but humans will not be replaced overnight because where empathy is needed, real people are still better than bots. In addition, many customers still want to talk to a person. A transition to more automation with AI is underway, but enabling humans to perform their job better is where we should focus our immediate attention. Customer service co-pilots can dramatically improve service, satisfaction, and productivity for agents.