Todd Pruzan, HBRWelcome to HBR Video Quick Take. I’m Todd Pruzan, senior editor of research and special projects at HBR. Harvard Business Review.
In a rapidly evolving business environment, artificial intelligence has become an essential tool for customer service, and some would argue that customer service leaders are being irresponsible if they’re not using AI agents. We’re going to hear from Brian Donahue, VP of Product at Intercom, about why that is. Thanks for joining us, Brian.
Brian Donahue, IntercomHey Todd, thanks for having me.
Todd Pruzan, HBRBrian, you make a strong case here that leaders who work without AI agents are irresponsible. Why do you make that case?
Brian Donahue, Intercom: Well, with all the hype going on right now, I think AI is going to change society, but we’re still waiting to see how it all plays out. I think it’s actually easy for leaders to fail to recognize that it’s already changing the customer service landscape.
AI agents provide enormous value today, right now. And we, as developers who understand the strengths and limitations of this technology, believe that not using it is a huge missed opportunity. It’s good for your customers and it’s good for your business. The economics of customer service are really changing. So to not recognize that we’re in this moment, I think, is to be left behind.
Todd Pruzan, HBR: So can you give us an example of how AI agents have impacted Intercom customers?
Brian Donahue, Intercom: Yes, that’s right. When we launched over a year ago, our average resolution rate was 32%. So we thought, enough is enough, we have to bring this to market. It’s a lot better than it was before. And that average rate has now gone up to 46%. It’s been a nice incremental increase. And it’s still going up.
We have customers with resolution rates over 80%. So all you have to do is tell your customer service leaders that an AI agent could probably answer almost half of their call volume, and they can do it 24/7, in any language. That’s why we think it’s basically irresponsible not to use an AI agent today.
Here’s a specific example from one of our customers, SaaS company RB2B: They doubled their user base in two months and expected their support team volume to double as well. However, because our AI agents actually addressed customer questions over email, their support team actually handled 45% fewer questions. RB2B says that a year from now, their support team of 2 agents will be a team of 20. This shows how dramatically agents are changing the customer service landscape.
Todd Pruzan, HBR: Yes, these are compelling numbers, but customer support isn’t just about efficiency. It’s about quality. Leaders care most about the quality of the customer experience they’re providing. How can you ensure that AI agents don’t degrade the quality of support your end customers receive?
Brian Donahue, Intercom: Yes, this is a really important point. This is the biggest concern we hear. There are a couple of things to think about here. The first is not to equate today’s AI agents with the chat bots of a few years ago. These agents can have human-like interactions because their conversational abilities have improved dramatically. These agents sound human. So I think most people intuitively understand this from their experience interacting with ChatGPT and similar things. That’s the first point.
Secondly, I think it’s important to think from the customer’s perspective. What customers care about most is that their problem gets resolved as quickly as possible. For most of what they want, they don’t care if it’s a human being who does it. They just want their problem resolved as quickly as possible.
AI agents offer a chance to find a solution as quickly as possible. This is a big change. If the AI agent can’t answer, it will escalate to a human. This escalation process needs to be transparent, flexible, and efficient.
That’s where the nuance and subtlety of how you respond comes in. Being able to have someone say, “Oh, that’s OK.” “Okay, you couldn’t answer my question, but I’ll get a human to help you now.” That point is clear. How that interaction plays out can actually be the key to whether the other person is annoyed or feels good. But ultimately, you’re trying to speed up that resolution as much as possible.
The third is accuracy. It has to be accurate. This was actually the biggest risk when this technology was announced. Can we reduce hallucinations? This was a big focus for us. We’ve actually been able to reduce the hallucination rate, the error rate, to around 0.01%, which is very significant.
So when we think about how this impacts customer experience, we need to remember the quality of human-like interaction first. Second, the possibility of immediate response and immediate resolution. Third, high accuracy, i.e. no mistakes. We believe these are the main reasons why you should feel comfortable using AI agents even for the highest customer satisfaction.
Todd Pruzan, HBR: Yeah, great summary. Brian, thank you for sharing your insights about AI agents and the role they play in providing responsible customer service.
Brian Donahue, Intercom: Thank you, Todd.
Todd Pruzan, HBRWe spoke with Brian Donahue, VP of Product at Intercom. Click the link below to learn more:
