HealthJoy founder and CEO Justin Holland spoke to PSFK founder and editor-in-chief Piers Fawkes about using chat interfaces to make healthcare a more human experience

According to PSFK's recent Customer Service Debrief, chatbots are becoming increasingly commonplace to deliver better consumer experience—especially in healthcare, where they serve as an accessible first touchpoint for patients and professionals.

Entrepreneur Justin Holland spoke to PSFK founder and editor-in-chief Piers Fawkes about his chat-based healthcare app HealthJoy, which connects employees to information about their benefits, and the advantages and limitations of the medium.

Piers: From a user point of view, what does HealthJoy do?

Justin: Effectively, you're chatting with Joy, our AI chatbot. It's a conversation that you're having back and forth—sometimes it's scripted, sometimes it's not. It's very similar to the experience you'd have with a chatbot on WhatsApp, Telegram or Facebook Messenger. We're utilizing understandable, conversational components that are ubiquitous across the space. We're making it very simple to have a freeform or scripted conversation with our AI.

Where does it work well, and where are the places where a human needs to jump in?

Chat itself is asynchronous, so when you give someone a conversation, there's not an expectation of instant back-and-forth communication. I'll give you an example: intake or triage with an agent. Picture that long questionnaire you have to fill out when you go to the ER. Imagine that being done in a conversation with an agent: It’s really disjointed, you're waiting 20 to 30 seconds for answers. You may never get to the end of the questions you need to ask before the user becomes too tired.

That's a classic scenario where there’s opportunity to do the intake quickly with a chatbot instead, where you’re getting answers very fast and not waiting for an agent response. That's a case where chat works very well. Obviously, the places where there's always going to be a challenge are specific cases with specific consumers. A very complex case involving a lot of different moving parts is not where we're focused right now.

It sounds like when you have outlier cases, the system is not ready to respond, and it hands the user over to a human being pretty quickly.

That's exactly right. Typically the user is passed off in about 20 to 30 seconds, as we find that we won't lose the customer if it's done in that time frame. If it takes two to three minutes, we lose 50 to 60% of our customers.

There is always going to be a human augmentation, at least for the next 10 years, unless the logarithmic expansion of AI becomes much easier to manage. Right now 80% of our experience is volleying back and forth between Joy and the consumer. We'll chip away at that 20%, and it's going be a multi-year project—it could take dozens of years or even decades to get that down. We will probably never get to 100%. I don't think there should be an expectation of that. I think it's OK to understand that there are limitations right now.

On the topic of chat, I see two points of view developing right now. I see articles and reports championing conversational technology. Then sometimes when I talk to people, I feel there's a little bit of chatbot fatigue. Obviously, with all the technologies, there's a hype factor as well.

We’re seeing AI-as-a-platform companies like Amazon, Google and Facebook develop a lot of core technologies that make it more useful. To me, conversation is literally just a different type of browser. Saying it in’t useful is like saying that there are too many different browsers, like Internet Explorer or Chrome. With pfconversation, you're just exploring and navigating content in a different manner.

More insights can be found in our Customer Service Debrief on building loyalty through conversational AI and chatbots.

Lead Image: Woman texting via Shutterstock