Mastering CX: how AI can improve customer service

Consumer frustrations with long call waiting times and unresponsive chatbots are harming customer service metrics. But generative AI can transform how answers are delivered

Customer service is no stranger to the rise of automation, from having to choose numbered call centre options to asking online chatbots for advice.

While many consumers cite frustrations with such systems that haven’t quite got it right, increasing numbers of incoming queries have left companies reliant on chatbots to reduce long waiting times for customers and employees alike.

For Girish Mathrubootham, CEO and founder of Freshworks, a company that creates AI-boosted business software, there is a better solution available – embedding generative AI to deliver quicker, simpler and more seamless customer service, without sacrificing user experience.

“If a business scales from serving thousands of customers to millions of customers, it is not sustainable to keep hiring more and more people to deal with level one customer service,” he explains. Enquiries regarding order tracking or cancellation are simple and can easily be dealt with by generative AI. 

“Businesses have always driven automation through self-help, but generative AI is a significant leap in what can be accomplished in customer service,” he explains. 

More accurate responses 

Mathrubootham suggests consumers aren’t bothered if problems are solved by AI or humans, as long as they get the right answer, and fast.

Through generative AI, automated customer service can now handle queries in near real-time, in multiple languages and through a two-way conversation, even when the discussion is complex. It can also decipher and understand audio, images or videos.

“The biggest achievement,” Mathrubootham says, “is having a multi-turn conversation with follow-up questions. This powerful technology produces a faster, more accurate and personalised response.”

A major timesaver for human colleagues

Generative AI’s role within customer service is to act as a co-pilot, working alongside humans to make their day-to-day tasks easier. This frees up human agents’ time to solve more complex problems or helps them complete administrative tasks more efficiently, such as locating information from huge digital knowledge bases.

If you put garbage in, you get garbage out

It can also proactively provide quality control on replies to customers, monitoring outbound messages and suggesting better responses.

“The AI might say ‘this doesn’t look like the right answer’,” Mathrubootham explains. “It can also detect if the tone isn’t professional or courteous, raising prompts to rephrase. This feature is particularly useful for training new employees.”

For generative AI to perform at its best, companies must initially feed in the right training data, set strong guardrails on the language used and ensure systems are secure.

One solution is to use the best and brightest human customer service agents in an organisation to train models, with these colleagues continually testing and refining automated responses until they sound human. 

“If you put garbage in, you get garbage out,” Mathrubootham warns. “When you’re training machines with data that is not accurate, machines can pick the wrong answers. You can’t just feed them a million customer service calls or a million response tickets from the past.”

The route to strong business metrics

Customer happiness is critical to long-term success and embedding generative AI within customer service is now key to achieving satisfaction at scale. 

For example, companies with high levels of unstructured data, such as customer satisfaction surveys or email queries, can quickly use it to pinpoint what support enquiries are most common. This allows organisations to highlight nuances from customer replies to clearly show why they are satisfied or disillusioned and increase productivity by dealing with simple queries in large batches. 

Implementing technology from a trusted partner can prove the best value. Freshworks’ Freddy AI, for example, uses its existing data training sets alongside a company’s internal databases, creating a tailored customer experience. 

“When we talk about empathy and a human experience, some of this is readily available with generative AI,” says Mathrubootham. “Customer service leaders must accept this is a big opportunity for their metrics, improving average handling and resolution times.”

Mathrubootham explains that over the past three or four decades, businesses had to use humans to structure data in CRM or helpdesk systems to access the best insights, a process which lacks efficiency and limits productivity. 

Now, generative AI is able to automatically complete these tasks in real-time, fundamentally changing the rules of the game. To unlock the benefits, businesses must pay attention to what their customers care about most and invest in a comprehensive generative AI partner.

For more information please visit