Health and dental plan company Simplyhealth is using generative artificial intelligence (GenAI) technology to cut the time it takes to respond to customer emails.
The company said its use of Salesforce’s Einstein GPT within the Denplan side of its business was helping it respond to customer emails around its three most frequently asked questions, such as, ‘How do I change dentist?’.
“We have consciously made the decision to take a test-and-learn approach to rolling out this capability, and we will grow our knowledge base slowly and carefully with the intention of incorporating more questions,” Simplyhealth’s customer service director Dan Eddie told Computer Weekly.
Previously, all emails were answered manually, with a response taking about 12 minutes. Since rolling out the AI tool, the team has been able to respond to these common questions within a minute and a half.
“Not only does this drive improved productivity, but it also provides more time for the team to resolve more complex questions and complete more value-adding work for our patients and customers,” said Eddie.
The GenAI tool answered around 10% of Simplyhealth emails in November and December 2023.
To ensure the AI is creating the right messages, Simplyhealth has built out a knowledge base using 500 previous email responses for its three most frequently asked questions. “This ensures that we have a strong and accurate information base for responses to be generated from,” said Eddie.
“We have consciously made the decision to take a test-and-learn approach to rolling out this capability, and we will grow our knowledge base slowly and carefully with the intention of incorporating more questions” Dan Eddie, Simplyhealth
The company also always has a human in the loop, so once a response is generated by Einstein GPT, this is always reviewed and approved by a member of the team to ensure the message is correct and accurate.
“We have been honest and transparent with our teams about the role that AI can play in our business. We are, and will continue to, adopting a test-and-learn approach, building up the use of AI in a controlled environment. This ground-breaking technology is about complementing the existing work our teams do,” said Eddie.
Simplyhealth now handles 35% of all customer traffic through live chat using Saleforce’s AI-powered Einstein bots. It is also using the Salesforce Service Cloud to speed up customer responses and enhance the accessibility of services. Call volume has reduced by nearly 40% and agents are now able to run numerous live chats simultaneously, according to Salesforce. Simplyhealth is also using Service Cloud Voice to produce real-time call transcriptions and summaries, which can be used as a “single source of truth” for every customer, helping inform agents ahead of future conversations.
Simplyhealth said it was taking a “controlled approach” to rolling out this AI capability, and the speed at which it adopts this technology will be governed by its in-house AI Forum, which comprises managers from various departments across the business. The next step is to increase the number of questions the AI can answer from three to five.
“As we strengthen our knowledge base within the system, we will then be able to review and explore opportunities with how we can incorporate the use of this technology within other digital contact channels,” said Eddie.
Research commissioned by Salesforce last year found that 90% of service professionals who use generative AI said it helps them deal with customers faster. However, it also found that sales and service workers are less likely to use GenAI tools than people working in marketing, perhaps because of a lack of confidence.
Healthcare is one of the areas that fans of GenAI think it can make the most impact, whether that’s by streamlining processes and saving time for doctors, such as by automatically summarising a consultation, or by creating new services.
Partly, that’s because there is a huge amount of data in healthcare – from medical trials through to patient records – that can be analysed to optimise services for patients; something that generative AI is very good at.
For example, GenAI is already being used in drug discovery or to analyse a patient’s medical history to offer a personalised care plan. The World Health Organization recently published guidance on the use of large language models in healthcare, reflecting the rapid growth in interest in this area.
There’s also an apparent appetite for more GenAI in healthcare generally. According to research by consultancy Capgemini last year, titled Why consumers love generative AI, 67% of consumers believe they could benefit from receiving medical advice from generative AI and 63% are excited about the possibilities of GenAI bringing faster and more accurate drug discovery and development.
While some of these use cases are still being developed, there are already examples of how GenAI is being used to deliver better services. For example, Amazon recently detailed how it is using GenAI to fill prescriptions more quickly and accurately at its Amazon Pharmacy service.
While doctors being renowned for having bad handwriting is less of a problem now, as prescriptions arrive at the pharmacy electronically, they can still contain confusing or inconsistent language, such as “take by mouth” or “take orally”, Amazon said.
To tackle this, Amazon ran the original unstructured data through a generative AI model which used “named entity recognition” to create a structure for the text, using categories such as “dose” and “frequency” which make it easier for clinical staff to fill prescriptions quickly.
Every prescription is still reviewed by a pharmacist in case of errors AI cannot detect – for example, sometimes the prescriber will put “volume” in the field meant for “strength”. Amazon said that by combining a GenAI approach with the expertise of a pharmacist, it can increase order processing speed by 90% and reduce the rate of human error.
It’s also using generative AI to help the Amazon Pharmacy clinical and customer care team answer questions faster by reviewing internal documentation pages and knowledge bases and summarising them in a “useful and clinically appropriate way”. Clinical and customer service reps then review everything before they speak with the customer.
