Creating better experiences for everyone with generative AI

This media and entertainment company prides itself on creating unrivaled customer experiences. With 24/7 customer service channels, these interactions present an opportunity for the company to understand its customers better.

In particular, the company knew that customer service chat conversations and app reviews hold a treasure trove of data.

But turning data into insights is often easier said than done. Text data is unstructured, meaning it doesn't fit well into a spreadsheet for quick analysis. Thankfully, new developments in machine learning – specifically natural language processing (NLP) – allow computers to interpret, manipulate, and comprehend human language so data scientists can automate much of the work.

Plus, the company didn't just want to analyze historical data – it also wanted to analyze customer chats in real time. The goal was to provide its customer chat agents with an ideal response using the power of generative AI (gen AI). But this would require customizing its own unique large language model (LLM).

The company recognized this would be a complex project – one it would need outside help with. Thanks to a longstanding relationship with Genpact, the company knew we were up to the task.

We broke the project into three steps:

1. Analyzing historical chat data: We used an open-source LLM to organize chat conversations by keywords, query, sentiment, and more. With these initial rules in place, we ran immense amounts of chat data through the LLM to categorize the data and customize the model. This step saved massive amounts of time compared to traditional, manual data analysis processes
2. App review analysis: With the app reviews, we needed to know what customers raved and ranted about so the company could create five-star experiences every time. To do this, we applied our learnings and process insights from the chat data review. Throughout steps one and two, we anonymized the data to protect customer privacy, in keeping with responsible AI best practices
3. Creating real-time chat suggestions: Finally, we took the analysis from the first two steps to create real-time response suggestions for chat agents using gen AI. This is where the real magic happens. Instead of just learning from the past, the company also gets insights into present interactions. Our automated chat recommendations help employees to answer even the most difficult questions in the company's brand voice

With any digital transformation, businesses want to be sure that the technology they use enhances human expertise. Thanks to gen AI, that was certainly the case here.

The company's chat agents can respond more quickly to customer queries by selecting responses generated by AI and then tweaking them if necessary. The company can also unlock insights from chat conversations and app reviews to inform future decisions.

Here are just some of the benefits:

• A better customer experience with personalized responses and recommendations
• Improved employee productivity because of automatically generated responses
• A consistent brand voice thanks to intelligent tone recognition
• Better vendor relationships as chat responses suggest vendor services customers might be interested in, bringing them more business
• Greater insight into customer relationships by understanding how customers interact in the app store
• Increased revenue as gen AI spots cross-sell and upsell recommendations for chat agents to act on

Of course, it's still early days for gen AI. But as this company and countless others are realizing, gen AI presents incredible opportunities to reinvent how people work.