How the Next Generation of Conversational AI Can Revolutionise your Contact Centre
With David Stone, Senior Vice President at Curious Thing AI
Show Notes
David Stone is a Senior Vice President at the conversational AI startup, Curious AI.
With 30 years’ experience in the call centre industry, he’s spent his career working in contact centres, for outsourcers, and for industry vendors.
Over the past 10 years, he’s seen the evolution of AI in contact centres. Today, he shares how the next generation of conversational AI can revolutionise your centres.
You'll Learn:
How David gets asked all the time by senior and CX leaders how their centres can take less calls from customers, and his unexpected answer (02:01).
The way conversational AI can be a bridge from your voice to your digital solution, at a lower price, and deliver better customer experience than an agent (03:11).
The 3Ps of this new wave of conversational AI, that allows for powerful customer conversations at scale (03:41).
Why traditional ‘conversational AI’ is neither conversational, nor overly intelligent, and why the next generation of the technology is vastly different (04:26).
How this next generation of the technology can be deployed within 10 days as opposed to months like traditional conversational AI, at a 10th of the cost (05:57).
Dave’s mantra for which companies will win in the digital age, and how you can join the winning side (08:21).
How you can tell when customers are likely to need to call your centre and proactively reach out to them, instead of needing to be able to take voice calls (08:48).
How conversational AI helps agents be even better, and will never fully replace their role (11:00).
How Curious Thing was able to able to deploy a solution for a government agency in Australia in 4 days, to make 1.5 million personalised calls (15:19).
How conversational AI is able to drive a constant loop of innovation, helping you gain better customer insights, and provide more value to them (17:07).
Transcript
Blair Stevenson (00:00)
Welcome to the Secrets to Contact Center Success podcast, connecting you with the latest and greatest tips from the best and the brightest minds in the industry.
I am Blair Stevenson, I'm the founder of BravaTrak. Our Sales Leadership System enables contact centres to increase revenue and achieve their sales growth targets.
Today I'm very fortunate to be joined by David Stone, who's a Senior Vice President with the conversational AI startup Curious Thing AI. So, Dave, welcome along. Great to have you.
David Stone (00:28)
Pleased to be here. Thanks for inviting me.
Blair Stevenson (00:30)
Fantastic. You're welcome. So as a starting point, just tell us a little bit about yourself and your experience.
David Stone (00:36)
Actually, when I think about it, I realize how old I'm getting.
I started my career in call centres about 30 years ago. I worked in an emergency management call centre for the London Ambulance Service. We did emergency calls and we supported disaster management. And it was there I fell in love with the notion of people, technology and process. And those three things coming together means that you can almost, there's no obstacle that you can't pass when those things sing in harmony.
And then I came to Australia about 25 years ago, and I worked in an outsourcer and I started on the phones, and eventually I ended up in a management position, went through the ranks. And along the way, vendors used to come in selling their wares. Things have changed now, but back then, they really weren't familiar with the outcomes we wanted and operational management. So I moved on to the vendor side.
So for the last 20 years or so, I've been in and around contact centre customer experience and AI technology is kind of what I've done.
Blair Stevenson (01:41)
Nice. It seems to me that conversational AI is now starting to show some really serious promise. But in terms of a use case for a conversational AI, how can it help contact centre leaders improve the customer service experience while reducing costs?
David Stone (02:01)
Yeah, it's funny. In my career, I continue to present to the C level, to boards, and to customer experience leaders. And I reckon 99% of the time in the break I get asked this question. It's like a dirty little secret it's, "Our customers keep calling. How do I take less calls?" And, you know, it's funny and it's like that. It's like, "Oh, you know, they keep calling me."
Firstly, I think it's great if your customers are calling, you're doing something right. I'd rather have that problem than the other one, right? But the problem isn't that they keep calling. The problem statement is wrong. The problem is you can't consume their chosen method of communication, which is voice.
Maybe it's no surprise, right? Because there's two things that impact your ability to do that, one is people. And people at the moment are really hard to find and really hard to retain. And particularly with onshoring. We're in a situation with near full employment in contact centers. And the other one is cost. You know, the cost of people is going up. And particularly when you onshore.
So certainly there's two main use cases for conversational AI. One is the ability to consume that voice. So someone's choosing to call you using their voice. Well, that's a bridge for you to digitize. So conversational AI is that bridge from the voice to your digital solution. It means you can do it at a lower price, it means you can do it faster, more efficiently, there's no waiting. But you can still do it with the same level of customer experience as if you had a person. So that's the first one.
The second use case, and one I'm seeing increasingly, is this notion I have of personalized, predictive and proactive. That is, the capacity to call out and reach out to your audience. And not just a small number, but a really large number, and engage them in a conversation.
So an example would be a client of ours in the US at key anniversaries, we reach out and engage and then pass that call back. So there's these two things, that capacity to be proactive, and that capacity to consume either, probably automate or partially automate, an inbound call.
Blair Stevenson (04:18)
Cool. Seems to me that traditional AI is not really delivering that yet. Would that be fair?
David Stone (04:26)
Well, look, I've got to be careful. I definitely don't want to knock the industry I'm in. But if I was honest, I would say a lot of conversational AI I see is neither conversational, nor is it overly intelligent.
And I understand why. Pioneers will always pay a price. There is a reason. They are actually constrained by design, Blair. Most of, if not all of them, except the new wave of conversational AI, is built on this notion of intent. So companies spend weeks and months before they deploy, trying to think of every single way someone might ask for something, what their intent might be.
That's slow, it's expensive, it's cumbersome and it's difficult to manage. And here's the really annoying thing; it's not a great experience.
An example would be, "Would you like an apple, an orange or a banana?" What if I want a pineapple? And it's typically, "Say after me, this." It's like no conversation I've ever had. It's actually like a voice version of the old IVR. It's very, very constrained. The world's kind of moved on from there, I think.
Blair Stevenson (05:46)
So I'm guessing that Curious Thing has moved on from there as well. What differentiates Curious Thing from what's gone before?
David Stone (05:57)
Curious Thing really is at the front of a new wave of doing conversational AI. The notion of intents has gone.
So I won't get into too much of the technology of it. But if you imagine an intent is a bit like being on a train. So once I've decided you want an orange, you're on this train and the first and only stop is orange, the only way you can get off is by jumping off that train or hanging up that phone. It's one of the challenges with the old intent way of doing business.
So Curious Thing doesn't use intents, we use a notion of topics and knowledge graphs. What does that mean to the non-technical like me? It means, instead of saying, "Do you want..." and "Say after the tone exactly the way I've said it, apple, orange or pear", we might say, "What fruit do you like, what fruit you want?" And you might go, "Well, I really like oranges, but today I want a banana."
Two things have happened, right? One is the customer is using their natural voice. Yes we've found out what they want, but we've discovered insight along the way. So what it means is, with Curious Thing and technologies that now use the way we use it, is you can have a completely open conversation. Very, very easy for your customer to say exactly what they want, the way they want to say it.
Because we don't use intents, it's very fast to deploy. So we would typically deploy a solution within 10 days as opposed to months. And the other thing, whether we like it or not, cost is always a factor. It's about a 10th of the cost of a traditional sort of conversational AI solution.
So all of those things are coming together. But fundamentally, the experience is so much better. So the capacity to inform and resolve is there, but also discover.
One of the other things is Curious Thing has really built a reputation on the capacity to do outbound, which most conversational AI solutions won't do.
Blair Stevenson (08:00)
Okay. So we'll come back to that. One of the things you mentioned earlier was this concept of personalized, predictive, and proactive. And I think personalized makes sense. Perhaps as a starting point, just talk us through what you mean by predictive and proactive.
David Stone (08:21)
I believe he who knows their customer best and acts fastest will win in the digital age. It's not enough to know them best, you've got to have the information and you've got to be able to act on it. Otherwise you're gonna lose. There's some great companies that are really good at having customer insight, but what do they do with it? And when do they act on it? You got to have it and you've got to be able to act on it.
And also, if you think about the single biggest problem I see, which is "I can't take voice calls", there is a way of not having your customers calling you; it's reaching out to them when you think they need to call you.
So an example, and someone said to me, "How could you possibly know?" And I said, "I'll tell you how you know", and I've experienced this. So you work for a telco. You've suddenly decided that you're going to change the look of the bill dramatically, and you're going to send it out on a Friday afternoon. Well, if you walk down to your contact center and you say to the agents;
"Hey, guess what guys? We haven't sold you this, because we don't, this is what we're sending out today. What's going to happen on Monday when we open for business?” And they'll say to you, "Well, between 12 and two at lunchtime, we're going to have an avalanche of calls. And they're going to be from this cohort about that bill."
So my notion of personalized, predictive, and proactive. And this is a really simple one, but there's so many, is why wouldn't you, if you know that, why wouldn't to that cohort do something as simple as maybe you send them an SMS and say, "Hey, Blair" Or if you know you prefer to call Mr. Stevenson, Mr. Stevenson, whatever it is, "Hey Blair, you're going to notice on Monday, we've got a new bill. If you click on this link, it'll take you to a YouTube video that I'll explain a bit about that bill. If you've got any questions, just send me a note on this SMS. My name's Dave and I'm your virtual assistant."
You're doing a few things. You're letting them know it's coming. Your inviting them to proactively understand the bill, which will call deflect. But even if they reach out, it's not actually a bad thing, because remember, these are people that we're going to call you anyway. And even if they reach out on the chat bot, you've suddenly got them on a digital channel. And if you do give them a good experience, they may never call you again, and be very, very happy.
And that's just an example. And I think what you'll see, one of the things I think will happen in our industry is inbound call centers will turn into outbound call centers. There'll be a disproportionate weighting to outbound, because we'll know.
Blair Stevenson (10:48)
That makes sense. But does that mean that's the end of people. Are we just going purely digital?
David Stone (11:00)
When I look at digital-first initiatives, when I look at customer experience and self-service initiatives, many of them miss one really important goal. And the goal is paradoxical to what they're trying to do.
What I should do in my job driving conversational AI, my number one goal shouldn't be to automate. It's part of it. It's to drive better conversations. So people won't go away because they're going to be needed.
When someone decides that they need to engage, it's usually a high value situation. Either, if it's government, it's someone who's really at risk or at some dramatic need, or it's a high value sale. So it definitely doesn't lead to the end of people. We can't get enough people in our industry, frankly, at the moment. What it means is it will take people to high value purposes.
So examples of that are a bit like I mentioned the customer in the US. Rather than have them all calling out indiscriminately to their customer base and engaging, and a large percentage don't want to be engaged, we reach out with conversational AI, they've got the option. They're not interested. We grab a call and we pass that call back, it's someone who wants to engage with the person. So we'll take them to higher value tasks.
Inbound calls, if it's a complex call and we're doing some work with government at the moment where before the call comes in, the client needs to be ready with certain material, we'll make sure when they call in, "Hey, do you have that? We can wait till you've got that", so that when they get through, it's a higher value call and a better conversation. So no. Real conversational AI will augment people. That’s what our job is.
Blair Stevenson (12:52)
Yep. That makes perfect sense. So people are still important, they deal with higher value calls, or they deal with people at a time where they have all the information they need together. So how can conversational AI support agents as they work?
David Stone (13:13)
There's a couple of ways, and I sort of touched on it there. Certainly one of them is this capacity to triage a person before they come through, to understand their need a little bit more and then pass a really - I hate the word 'qualified' because it's not just not about sales - but a really contextualized call will come through. That when the person comes through that we are understanding what they want, how they want it, and what need we haven't been able to fulfill. That's certainly one way.
The other way that I think is coming. And I say, I think, I believe it will come. I see a future where conversational AI is going to be sitting like your perfect coach and the capacity for it to prompt you, either verbally in your ear quietly, where the caller can't hear it, or on the screen.
And that's really important because I'm still sort of puzzled why it happens, agents have to jump in and out of multiple applications, they have to remember a whole lot of complexity and compliance. So I think in those two ways, by again, enabling that better conversation by pre-arming people, by giving them some time and making sure that the people coming through are knowledgeable about what they're going to talk about, and it's really clear.
And down the track, I think it will be almost like conversational AI meets knowledge management.
Blair Stevenson (14:46)
That's powerful stuff.
David Stone (14:50)
I definitely think it's coming. And conversation AI, we've just scratched the surface. And as I say, the thing Curious Thing is doing, it's so far away from the traditional in terms of speed and pricing and capability of complexity, but it's going to go again, because they’re really, really smart people.
Blair Stevenson (15:11)
So what's an example of a current client application where Curious Thing is helping out?
David Stone (15:19)
Something that's quite topical. So for a government agency, there was a need to let an awful lot of small to medium businesses understand, not only understand what their obligations were for compliance, but understand what was needed to be done to comply, to impart that knowledge, "Do you understand what you need to do? This is what you need to do."
And it needed to happen very quickly, because it's all about minimizing the spread of COVID, but also enabling us all - we're all in lockdown down at the moment - enabling us to open. So for this client, we identified that what was needed was about 1.5 million calls.
And you're saying, "We want to make about 1.5 million calls. We want to do that really quickly. We want to do it really cost effectively. Can we do it with people?" Yeah, we can, but we've got to go find them. We've got to recruit them. We've got to train them. We've got to find them somewhere to sit. We've got to give them technology. We've got to put in privacy. We've got to do all of that stuff.
Well, within four days, Curious Thing guys won't thank me for saying it because they don't like doing it that fast, we would have rather have done it in eight days. But within four days we deployed a solution.
The solution reached out to up to 70,000 people a day. It had a personalized conversation. No two conversations may be the same using real conversational AI. Every conversation could be different. We will ask the question, "Do you understand what you need to do?" The answer could be myriad, right? And we would engage them in a conversation, ensure that they understood what they needed to do, where they needed to go.
Some of those people we would transfer to humans, remember that whole thing of taking humans to a high purpose? Rather than ringing people up, "Do you know what you need to do?", "Yip." Well those that don't, we had people free to take those calls. So they were routed through to the government people to engage them in conversation.
We did 1.5 million calls over that period. The peak we did a hundred thousand calls a day. What was also very important in that wasn't just that resolving piece, not just that transferring piece, but government could look at the insight. What were people saying to us? Because we give a full transcription of the call. They can understand, "Is my messaging being received? Do I need to change my messaging? Oh, people didn't understand that bit." So again, that insight drove a better customer experience, that constant loop of innovation.
Blair Stevenson (17:37)
Wow. That's impressive. One last question, which really comes back to this idea, you talked earlier about how you believe there's going to be a shift from predominantly inbound to predominantly outbound.
So I'm just thinking in terms of the technology as to where it is now, and where it might be by mid this decade, in perhaps another three or four years time, where do you think it will be? When do you think that kind of shift from inbound to outbound is going to start using conversational AI?
David Stone (18:10):
I think it's definitely started. We're seeing it with some of our particularly FinTech customers, who are actually traditionally light on people by the notion of being a FinTech. But even more so, we're supporting the smaller number of people they would have as a percentage of customers to most, we enable them to have better conversations.
So the shift has started to this "I need to know my customer better." Everyone knows, everyone's busy now trying to, 'customer insight' is a bit of a buzzword. But then being able to translate that on scale is the next challenge, because, "Well, I've got the insight. How do I possibly reach out to all these people? Why would I? I don't have enough call center agents?"
Well, actually you can use conversational AI. And in fact, for some of that customers, and that government customer, we could have called all those million people probably in two days, but they couldn't consume the escalated calls. So it's started to happen.
I think the tipping point, I'll be bold and say, four years, five years. And does that seem like a short period of time? Not really. Not when you think of how far self-service and digitization have come, a year of massively reduced call volumes while trying to maintain service.
So this piece, the learnings from there will apply here, and it will be driven more and more. So it's already happening. And those that don't will be laggards. And I think what will drive it is the increasing acceptance. And I have to thank Siri and all of that, because people are quite comfortable talking to machines now.
They just don't like the IVR process. But the notion of saying, "Just tell me what you want." And when I demonstrate the solution to a customer and say, "Ask it a question." And they're self learning as well. I think that's the other thing.
Unsupervised self-learning is an opportunity and a risk, because the system gets smarter every time. So from history for a good conversational AI is the last call, how was that question asked on the last call? How could it be made better?
Blair Stevenson (20:17)
Wow, that's impressive stuff.
David Stone (20:21)
And it's coming along the way very, very quickly. And the use case is almost limited by your imagination. I mean, we're doing things now that are of real social importance, that just couldn't be done before, because you didn't have the volume of people, I'm talking about some of the government workers.
Blair Stevenson (20:39)
Yeah. Dave, thank you so much for sharing your experiences. Fascinating stuff. Appreciate it.
David Stone (20:47)
I really enjoyed it, thank you, Blair.
Blair Stevenson (20:49)
For listeners, you'll find the link to the show notes in the episode description below.
And if you'd like to connect with David on LinkedIn, you'll also find the link to his LinkedIn profile in the description too.
If you'd like to follow me on LinkedIn, you'll find the link to my profile there as well. Well, that's it from us today. Have a productive week.