Humanlike AI agents must master three core skills: listening, reasoning, and speaking Read more

How Pacific Gas and Electric reduced customer effort by 25% with PolyAI

Industry

Utilities

Use Cases

  • Authentication, call routing, outages, billing, FAQs (including Spanish)

Integrations

  • Cisco, Oracle, Nuance, Kubra, Mulesoft

Featuring

Kristin Punter
Sr. Director CSO and Strategic WFM
Pacific Gas & Electric

Pacific Gas and Electric Company (PG&E) is a utility provider headquartered at Kaiser Center, in Oakland, California. As California’s largest utility company, PG&E provides natural gas and electricity to 5.2 million households in the northern two-thirds of California.

Challenges

Customers being forced to engage with an outdated IVR system that was unable to meet their needs, leading call wait times of over an hour during due to unpredictable weather & natural events

Solution

An enterprise generative deployment that meets strict regulations, authenticates customers to keep accounts secure, scales to meet demand during outages and handles customer queries in multiple languages

Result

Over 4 million calls are resolved each year by PolyAI agent, increasing CSAT by 22% during outages and reducing overall customer effort by 25% compared previous IVR system


 

Challenge

PG&E received approximately 16 million calls a year but unpredictable weather events caused extreme peaks in volumes, with daily call numbers in the tens of thousands during these emergencies. These problems were compounded further by call center agent retention challenges and an outdated IVR solution, leaving customers frustrated and waiting over an hour to speak to someone.

The existing IVR was built by their CCaaS provider, who had automated common call flows like payments. However, managing the system proved difficult and costly. With over 7,000 utterances in the IVR, updates were time-consuming and expensive, limiting the team’s ability to enhance the customer experience over the phone.

Additionally, PG&E was about to implement a rate increase that would create a substantial surge in call volumes, putting additional pressure on the existing system, and ultimately lengthening wait times even further. The team needed a solution that could hit the ground running, scale quickly to meet demand and satisfy strict industry regulation and safety requirements, all while providing a positive customer experience.


 

Solution

PG&E partnered with PolyAI to build an AI agent that could initially work in conjunction with their existing IVR. This approach, as well as starting without any API integrations, allowed the team to get a solution up and running quickly without rebuilding their existing contact center infrastructure. With this initial deployment success, integrations were built into key platforms like Oracle and Cisco, as well as numerous in-house systems. These integrations have allowed the scope of use cases handled by the PolyAI agent to increase dramatically.

Distressed customers calling the dedicated outages line are immediately greeted and reassured by PolyAI’s agent, Peggy, that the situation is under control and are provided with an outages update based on their location. Peggy has the ability to give information about the outage, including an estimate of when the issue will be resolved with accuracy and compassion.

Additionally, Peggy answers billing inquiries and FAQs in both English and Spanish, with the ability to text customers a link to specific online resources where they can self-serve without speaking to an agent, further reducing the volume of calls that go to the contact center.

With utilities being a highly regulated industry, it was essential that this solution met strict regulations and safety requirements. The AI agent handles millions of calls a year across highly sensitive topics, including potential gas leaks, where any wrong answer or error can have disastrous consequences. PolyAI’s robust guardrails and proprietary tech stack ensure queries are answered correctly and customers are accurately routed when needed, enabling Peggy to be the first generative deployment in a regulated industry.

Results

22%

Increase in CSAT during outages

25%

Reduction in customer effort

41%

Of all calls are resolved by PolyAI

PolyAI worked with PG&E to establish a positive ROI target of 3.5% call containment. PolyAI’s agent surpassed that on the first day, successfully containing 15% of calls, and now averages a 41% successful self-service rate across all calls. As new use cases are identified, this number will continue to go up. Customers have been surprised to have their calls answered immediately and have been thanking the AI agent for its help, which is reflected in a 22% increase in CSAT for outage calls despite this being a traditionally distressing call type.

Furthermore, the ability of the PolyAI agent to understand and speak in complete sentences allowed it to achieve a 25% reduction in customer effort via more efficient calls compared to the previous conversational IVR solution. Additionally, PG&E previously had very little insight into its call data and call types due to mis-routing and unreliable after-call notes. PolyAI accurately records call data, giving the company a much clearer understanding of their customer needs and empowering them to make informed decisions for future automation.

Future plans

Having achieved significant success so far, the PG&E and PolyAI teams are continuing to build and expand the scope of the AI agent. Start/Stop service, appointment setting and expanded billing capabilities are just some use cases being built in the coming months, with each set to be available in English and Spanish satisfying PG&E’s diverse customer base.

With these use cases in place and 100% of calls handled by the Poly AI agent, PG&E aim to sunset their legacy IVR by the end of the year.

Ready to hear it for yourself?

Get a personalized demo to learn how PolyAI can help you
 drive measurable business value.

Request a demo

Request a demo