This week on the Friday 15:

– Amazon’s bigger investment in Anthropic and what it means for the future of eCommerce

– A new AI chatbot tool called Hume and how it’s going to affect first-line customer service workers in the future. Plus, Andy shares and example of an interaction he had with Hume…

— 

Brian Beck: Andy Hoar, welcome to the Friday 15 with Master B2B. It is Friday, April 5th. I’m excited to be back with you today. Andy Hoar and I are here, my partner in Master B2B, our thought leadership series. This is Friday 15. We’ve got a jam packed and very, very interesting topic today, Andy. Talking about AI and where it’s going to be applied and B2B commerce. In this case, we’re talking about customer service. So excited to get into our topic today. And of course, as always, we start with some… breaking news. We didn’t know if you saw this this week, but Amazon doubled down on its investment in Anthropic, completing its planned $4 billion investment. This is huge. They’re placing a huge bet here, competing with Google, with OpenAI, Microsoft. So they’re doubling down here, Andy. Any thoughts?

Andy Hoar: Yeah, it’s game on now. First of all, it’s a big sum of money. And the fact that they’re, again, doubling down on it. And like you said, Microsoft, Google, Amazon. And by the way, Apple is rumored to be close to revealing some of their strategy around AI. So those are the big four. Of course, we’re going to have Meta and others. It’s going to be at least four or five companies that are going to be competing in the space, which makes for an interesting question about what is AI going to become? Or we’re going to have versions of AI. We’re going to have AI that works better for you versus for me. This is all to be determined.

Brian Beck: What’s fascinating to me is that Amazon, obviously, is a commerce business. So I think we’re going to see a lot of our applications emerging with AI. For the commerce side of things is with what Amazon is doing with Anthropic. Now we’ll see. I think we’re going to have a series of different options here that companies have available to them in terms of taking advantage of AI. And we’re going to be talking AI next week at our roundtable in Dallas. So I’m excited about what the industry has to say and continue the conversation there. But let’s get into our topic today, which is, will AI-driven chat bots, such as something called, “Hume AI,” which we’ll show in a moment, replace first-level customer service reps in B2B commerce? And first level, Andy, these are questions, right? Think about what does first level mean? It’s the questions that people come in and they ask and they’re looking for a quick answer. Or they’re looking for some level of service which may not be as advanced or complex as asking application or configuration questions. And it seems to me that we may have, we may be there already. And I know you looked at this Hume tool. Tell us first what this is. What is Hume?

Andy hoar: Well chatbots are not new. They’ve been around for a while and they’ve been, some have been good, some have been bad. I remember looking at one like 10 years ago that I thought was really impressive because I couldn’t tell whether it was a human or software and I thought, wow, that blew me away. Well, I was, I’ve been blown away again. There’s this product called Hume that is the first – that I know of – empathetic AI voice. And it’s interesting because what it does, if you go to the next slide, it actually analyzes your voice as you’re asking a question. You can’t really see it very well. People on the podcast, there’s a box on the side that shows a chart where it’s taking your voice, breaking it down and trying to determine whether you’re angry or happy. And then it actually answers based on its analysis of your voice in real time. It’s very conversational. And when we saw this, we thought, wait a minute. So you’re point about first of a customer service- How often have you gotten on the phone or been in a chat or even tried to text somebody and been stuck? You’re next in queue or you’re 18th in queue, right? I’m waiting for software. With something like this, I don’t think this is going to be a problem. This is infinitely scalable, at least in theory. So we wanted to show the audience how this works. So we did a little recording. It was a recording of interacting with Hume so you can get a sense for it. So:

[Hume Recording Plays]

Andy Hoar: Notice how all I did was ask a question. I tried to express frustration. It analyzed my voice and figured that out because it replied saying, “Oh, that’s a bummer.” “Oh, I’m sorry.” And it gave very practical advice. Now, just imagine a straight line between this and the future. How long before this is going to be first level customer service? You’ll contact a company, and it will ask you, “Hey, what’s going on? What’s the challenge?” And be able to process what you’re saying and read your voice and respond accordingly. So if somebody is really upset, you don’t say things like, “Hey, good to hear you today.” And if somebody is really excited, you can go, “Oh, that’s a bummer.” I’m sorry to hear about that. It seems pretty good and seems it sort of figured it out. Now, that was a practical use case. And I thought I would try something else. So if you want to see another video that I did here.

[Hume Recording Plays]

Andy Hoar: There you go, Brian. I’m here if you need to talk to me.

Brian Beck: This reminds me of “Blade Runner” or “Total Recall” or one of those movies way back when it’s more of like a buddy. Like you go out into your house and say, hey, how are you doing? How was your day? This is an empathetic voice. I mean, it’s a little too Southern California, right? But all that aside and obviously there’s processing time delays there. So it’s really trying to understand what the tone is. But I have to say that is incredible because if you just listen to how it responded and changed as you changed the approach. The problem was it was with MY team. It’s with my friend. Oh, I get it. Okay. Right? So I can see where this is going. This can be really incredibly powerful and almost almost a little disturbing, right? If you think about the applications 40 or 50 years from now we’re heading to dystopia. But anyway, let that aside, let’s get back to customers and what can we use this for with customer service and chat bots. They’ve been used for a while. This is a huge market, Andy.

Andy Hoar: I think some of this research was done before the release of Hume and some of these AI enhanced bots. But the predictions were that by this year we’d hit about a $7 billion mark. And I think that’s fairly accurate. Any of us who use customer service on sites, there are robotic chat bots powering those things, but they’re going to seem like child’s play compared to an enhanced version of say, Hume that is powered by and informed by all the knowledge that, can just be loaded in in different combinations, questions, answers. There’s some time associated with that. But imagine this for a moment when you bring a person on board and you train them, it takes a period of time to train them and the person can leave. And they only work certain hours of the day and it’s difficult work. With Hume, for example, you can train it once, you can keep training it. And it’ll never forget what you trained it. It’ll work 24/7, 365. And it’ll get better and better and better. And I think it’s pretty good at this point. So, even these predictions that by 2028, or 2029, it’ll be a $20 billion business. I think those are low. We’ve hit an inflection point. I think if these chatbots can actually be as good as what we just heard, you’re going to see these numbers go much higher than that.

Brian Beck: It’s incredible because I think this has implications for efficiency, for customer satisfaction, and also just for jobs. I mean, think about it. You’ve got call centers full of people doing first level. But what’s interesting here too is how do people use chatbots? People are not afraid to use them. The data is clear here. 88% of users engaged at least one conversation with a chatbot. Only, what, 9% of consumers oppose companies using bots. Millennials use them all the time. 40% engaged daily with digital assistance. This is becoming part of our behavior. We’re not afraid to use them. But there’s reasons for this. And it’s really about convenience, isn’t it Andy?

Andy Hoar: Yeah, it’s all about convenience. In some research that was done recently, a majority of customers said they would use an online chatbot instead of waiting for a customer service representative. And well, that’s just a perennial problem. We’re always waiting for customer service representatives and then increasingly disappointed with what they tell us. Two thirds of people said, look, on the margin, I’ll talk to a chatbot. So I think though, the thing that’s held the chatbot’s back is exactly what Hume seems to be addressing, which is – when you talk to a chatbot you know you’re talking to a chatbot, and it’s very transactional and functional. And the answers are just informational answers, which is why this last piece of research is particularly interesting. So they said “what would you personally use a chatbot for?” And there’s a series of maybe eight or ten answers here for the people on the podcast, the things like asking a question, making a complaint, making a purchase. Things like that. And what you see is the only stuff that’s above 50% is “finding a human,” which is funny. It’s good for asking a question, but not going very far down. I think the reason why is because we don’t really trust them. We don’t think they understand us. It’s a little too robotic. When there’s a human we assume they understand our frustrations. If you have a technology like this, I think it could bridge that gap and bring some of these other ones like making a complaint or making a purchase. These are things that can now go from being below 50% to much higher. And again, if it’s theoretically infinitely scalable and it really does a good job of understanding and empathizing, this could be a real game changer for customer service. There’s a point where honestly, why would you talk to a human anymore? A human who’s got information, maybe is having a bad day, him or herself. Why not talk to the robot?

Brian Beck: The way that the robot, the AI interacted with you, if you take away the lag time, it actually would sound very much like a real conversation. Now it’s still clear it’s something’s processing. You’re waiting two, three seconds for it to answer you. But that’s going to change. That’ll speed up. And you know, what’s fascinating to me about the data we’re looking at here, the number one, “what did you use a chatbot for?” It was finding a human customer service agent. And I agree with you. And at some point, these tools are more human and will reach a point where it’s indistinguishable. You don’t know you’re not talking to a human. There’ll probably be clues, you’ll suspect, but as the technology continues to get better, it’ll be able to just handle things like an escalation. “Oh, you have an application question, sir. I’m sorry. Can you hold for a supervisor, please, or our technical expert will help you?” So it’ll be a natural sort of progression through customer service.

Andy Hoar: Do you think that you think that the chatbot escalation will have a different voice? Do you think if we talked to the first level it’ll be a younger voice? And then when we refer this to my manager and the manager will sound like an older person. One thing we didn’t talk about, which just occurred to me is, on top of this – imagine when, not only can it handle you in real time, but actually connect it with past interactions. What if it welcomes you back? “Hey, Brian, good to hear your voice again. How did that thing work out for you?” “I know you bought this – did you have fun with your jetski?” it’s like, “what seems to be the problem?” “Yeah, well, it’s broken.” “Oh, really? Wow.” And that is, again, that is very easy for something like this to do and just think how connecting those dots would take this to the next level.

Brian Beck: We asked our audience, will AI driven chatbots, such as Hume replace first level customer service reps in B2B e-commerce? Oh, guess what? 80% said yes, first level goes to AI. It begs the question, Andy, if that’s true, then when is it actually going to be ready for that? And how should companies be investing in it? There’s opportunities for pure plays here, like Hume and others to come in, solution providers, as opportunities for Amazon, like we talked about in others, to create solutions here.