Customer at the Heart, Powered by AI

Delivering a “10 out of 10” experience means continuously evolving how we serve our customers, and AI presents a powerful opportunity to do that at scale.  

 
Today, at PG&E, we are intentionally leveraging AI tools to help our customer service representatives work smarter, respond faster, and create more seamless, empathetic experiences for customers. 

Leveraging the potential of AI

We analyzed the daily work of our customer service representatives and found two major pain points:

1. Searching for answers took too long. Customer service representatives had to look through a system of hundreds of procedures to find the right steps to answer questions. This was time consuming for them. 
2. Manually reading call transcripts. We had to manually read every transcript from calls to identify patterns that slowed our ability to identify reoccurring issues and opportunities.  

These two ideas became our first AI tools for the Contact Center: Agent Assist and Call Tagging. Agent Assist, which rolled out in April, provides real-time guidance and suggested responses to support customer service agents during interactions. Launched late last year, call tagging automatically labels and categorizes calls based on their content for analysis and reporting.

 
Building a safe and responsible AI foundation

Before bringing AI to our teams, we made sure it met a high bar for accuracy, speed, and stability so customer service representatives could rely on it every day. We also evaluated several partners and chose Amazon Web Services for its ability to help us quickly build, scale, and support the tool responsibly.

Testing the AI tools with CSRs

The early results were good. Calls got faster, answers were accurate, and customer service representatives said the tool helped them significantly. Newer reps asked for more step‑by‑step help, so we also added that.

Because the results stayed strong, we decided to use the tool more widely — it has been introduced to all 900 call center agents. We still collect feedback every month, have a 90% response rate on employee feedback surveys and keep a list of improvements to make. This helps the tool stay useful as our work changes.

Results and what’s next

AI Agent Assist is already delivering measurable value: reducing call time by 4%, lowering costs, and creating a faster, more effective way for us to test ideas, learn, and improve. Just as importantly, it has helped us strengthen our approach to using AI safely and responsibly while laying the groundwork to expand AI across more parts of the company.

AI Agent Assist didn’t just make one process better; it helped us build a new way of working. It showed us how listening to customers and coworkers, using data, and improving step by step can help us deliver “10 out of 10” experiences every day.

About the author

Vincent Davis is chief customer officer and senior vice president of Customer Experience.