When a hurricane hits, the last thing homeowners (or their insurers) need, or want, is to be on hold for hours to submit a claim. But with call volumes dramatically exceeding norms, that’s always been inevitable. Until now.
In 2018, within less than a week, Hurricane Michael suddenly became a major Category 5 storm and one of the largest storms to hit the Florida Panhandle since 1992. That’s when Security First Insurance, one of the state’s largest insurance carriers and a CodeObjects client, changed plans. Rather than pilot their new AI-powered virtual agent on a small scale, they decided to fully launch it to automate questions, requests and transactions, and better respond to the expected influx of P&C claims.
The team had a short window and a lot to accomplish. “It was all hands on deck,” says Arun Bala, Vice President of Products for CodeObjects.
Meeting high demand with frictionless AI-assisted customer service
The resulting AI-powered virtual assistant, InsurBot, is purpose-built for the insurance industry — and for their customers and employees. “We have prebuilt intents relevant to the insurance space, everything from policy sales and servicing to billing questions and transactions, to claims management,” explains Bala.
Additionally, InsurBot is multilingual and omnichannel, supporting voice, text, web and home automation devices with built-in analytics to guide constant improvement. By taking care of repetitive predictable transactions, like making payments, filing claims, getting copies of ID cards, and so on, InsurBot frees up employees to do more complex tasks. When customers need a human agent, InsurBot makes a live transfer. This transfer includes full conversation context, so the agent who takes over the conversation knows exactly what’s going on without requiring the customer to repeat information.
InsurBot employs other capabilities to make calls more frictionless, personal and efficient. “We’re able to identify the caller based on the phone number registered with their policy, which allows us to personally greet the customer,” says Bala. “We can also locate data in the system to help predict why they might be calling, and the likely conversational intents, to save time and speed response.”
From a platform perspective, Bala adds, “We did proof of concept implementations with many platforms. We found that for providing a superior caller experience and customer experience, the Watson technology and platform stood out clearly above the rest.”
For CodeObjects, Watson Natural Language Understanding is a key component in InsurBot’s success, ensuring the greatest possible levels of understanding for the bot. Additionally, Bala says Watson Assistant for Voice Interaction helps reduce the orchestration burden for his developer team.
Results that speak for themselves
In the wake of Hurricane Michael, 75 percent of the policyholders who chose to use InsurBot were able to complete business without being diverted to a live agent. All of the customers who opted for InsurBot experienced zero wait time, compared to an average wait time of 8 minutes, and as long as 30 minutes, following a hurricane the prior year. In all, InsurBot handled 30 percent of all the calls Security First Insurance received during Hurricane Michael, resulting in savings of approximately $1 per minute during the post-storm call spike.