ylliX - Online Advertising Network
Off Script: Imagining the next era of customer service

Off Script: Imagining the next era of customer service


AI is revolutionizing customer service as we know it, and businesses that seize this moment will lead the charge into the future.

Let’s put it this way: when we first got into this business in 2011, online customer service wasn’t exactly thrilling. The landscape was a bit stagnant, there was very little energy, and plenty of companies treated customer service as an afterthought.

But everything changed with the arrival of ChatGPT in late 2022. Advances in generative AI technology have unlocked a world of possibilities for customer service – and with them, significant challenges.

“Some say predicting the future is a fool’s errand, but we believe innovation thrives on imagination”

We knew it was crucial to address this head-on, so earlier this year, we launched Off Script, a series of candid conversations with Intercom leaders about this extraordinary shift driven by AI.

Some say predicting the future is a fool’s errand, but we believe innovation thrives on imagination. Throughout history, envisioning possibilities and embracing creativity has led to technology that profoundly shaped our lives. In a way, Off Script has been the ultimate exercise in imagination – a platform to share ideas, even the most provocative ones, so we can build a future of better experiences together.

In past episodes, we’ve discussed the evolution of machine learning, the changes we’re seeing in the software space, product design and customer experiences, and even how this is impacting the ways we measure and manage customer service.

In the final episode of Off Script’s first season, we wanted to look even further ahead. Here, Intercom’s Co-founder and CEO, Eoghan McCabe, shares his vision of customer service in an AI-first world.

Here are some key takeaways from the episode:

  • The AI landscape, especially in customer service, is now more competitive than ever, with success depending on who can execute and deliver solutions first.
  • Customer service bots are evolving from basic question-and-answer tools to more personalized, proactive agents capable of handling complex inquiries across multiple channels.
  • While AI is advancing rapidly, full-scale, high-performance AI-driven customer service is still likely several years away.
  • The resolution rate of AI agents is expected to rise as AI models advance and become more capable of handling complex inquiries and taking action.
  • The future of customer service will blend AI and humans to create better experiences, with human connections possibly becoming even more valuable.
  • AI is disrupting traditional business models. To stay relevant, businesses must adapt and innovate with AI-driven solutions.

You can find the entire Off Script series right here or on YouTube

What follows is a lightly edited transcript of the episode.


Off Script: Episode 6
Eoghan McCabe on imagining the next era of customer service

Des Traynor: This is the last episode in the first season of what we call Off Script. We’ve had our Intercom leaders talk about the future of AI, where the broad category of technology is going to take business, and take customer service.

In the first episode, I spoke about how AI is just going to fundamentally change software.

“It’s like an extinction moment. It’s like a meteorite hits your industry.”

And then after that, Paul, our head of product, detailed how it’ll change customer service software.

“I think there’s a very good chance it’s as big as the Industrial Revolution.”

Our head of AI, Fergal, then talked through the landscape of machine learning and spoke about all the new capabilities the world has.

Certainly we know many things that are possible now that weren’t before.”

Emmett, who’s our head of design, covered how the design and the production of software are just going to change forevermore.

There’s a lot we can learn from looking back, but there’s also the sense that a lot might be different this time around.”

And then in our most recent episode, our president Archana explained how all these changes affect how you measure and how you manage customer service.

Really, the question is how much do you care about your customers?”

So today, I’m really excited to introduce Eoghan, our CEO and one of our founders, to talk about where all of this might be going.

“A lot of this is fantasy. A lot of this is crazy. But a lot of it is also grounded in very real possibilities”

Eoghan projects a few years into the future to imagine what customer service could look like in an AI-first world. He explores an array of possibilities, from avatars to expert agents to bots talking to bots and beyond, and he asks some of the bigger questions: How good can this get? How quick can this happen? What role remains for humans in the whole process? And much more.

Now, look, a lot of this is fantasy. A lot of this is crazy. But a lot of it’s also grounded in very real possibilities. The world is moving so fast right now that starting to think about where it’s going is very helpful for those of us who are building here today.

Good service matters

Eoghan McCabe: What’s interesting now is that I think we’re now asleep to just how poor service online is. The opportunity was not just an obvious technology opportunity, but a giant business opportunity. In the year 2024, there is more capital and talent focused on the same set of problems than there has ever been before.

When we got started in 2011, it was just less competitive. You build something cool and you win, give or take. Now, in software, particularly AI, particularly AI for customer service, there’s just so much money and so much human intelligence applied to solving problems.

“Customer service is so insanely valuable, yet it’s not the thing we pay for”

Everyone’s going to figure out all of this stuff eventually, and what’ll matter is who’s able to actually build it and ship it first. So now we’ve got us and the big guys and the little guys all trying to reinvent themselves to become no longer software selling seats to humans doing the work, but software doing the work itself.

Customer service is so insanely valuable, yet it’s not the thing we pay for. The thing we pay for is on the receipt or the invoice, and the customer service is everything that is wrapped around that.

“If I see two doctors that are functionally the same, but one has reviews saying they’re kind and empathetic and thoughtful, I’ll pick that guy”

We know fundamentally how important it is. Think about how we talk about the bedside manner of a doctor. This doctor may be very well-studied, may have written many papers, attends all the conferences, but if they don’t have the soft skills required to connect with, empathize with, understand, and build trust with the patient, their ability to help them and their potential for success is going to be diminished.

If I see two doctors that are functionally the same in terms of their abilities and understanding, but one has reviews saying they’re kind and empathetic and thoughtful, I’ll pick that guy. The same is true with any kind of business.

The wait is over

Internet service is particularly interesting because when businesses started to move online, they did so to benefit from the great efficiencies of scale that the internet brings. You could sell around the clock, around the globe, and you disconnected from the physical, geographical, biological limits that real-world businesses had.

Inherent in that was the reality that any given business would have so much more customers than a real-world business would have – and almost the expectation was that you weren’t going to get great service. I think that the benefits of internet business made us all okay with that.

“If we can’t get an answer to the question we have when we’re trying to do something the business wants us to do, we’ll move on”

According to some survey I read, the average response time for an online customer service request was 12 hours, but we all know that it’s not untypical to have to wait days. That’s not okay. We are all so time poor as consumers, if we can’t get an answer to the question we have when we’re trying to do something the business wants us to do, we’ll move on.

What’s really interesting about this time is how AI is going to finally change that. I think that AI is going to allow us to provide service at a scale and to a degree of quality and speed that has not been possible heretofore, and that similarly will shatter expectations for forever.

The story of customer service online is a really short one. The internet as we know it is 30 years or less old. When businesses started to come online in the early 2000s, service was largely supplied by email. That had its limitations, and then companies like Zendesk showed up and allowed these businesses to manage the operations of lots of consumers and lots of support reps working together.

“It was super stale and stagnant until a year ago when everyone started to respond to this new AI world”

Today, we have many companies doing that category. It’s chiefly help desks with a lot of trimmings around it. You’ve got perhaps ServiceNow at the very, very top; Service Cloud from Salesforce below them; Zendesk below them doing mid-market; and then below Zendesk, you’ve got Intercom; and below Intercom, you’ve got a long list of small companies too. So you’ve got all these companies doing much of the same thing, which is managing conversations and tickets. Intercom was always a little different in that it was all about conversational approaches to support rather than email, which is what Zendesk did.

Our kind of insight or perspective on all this was that there was very little action there. There was very little change there. Very little energy. It hadn’t really moved or evolved in many years. It was super stale and stagnant until a year ago when everyone started to respond to this new AI world.

Not your average bot

The standard right now is that you’ve got these agents, and you can think of these as question-and-answer bots. Consumers give us questions, and we deliver answers as best we can. When we think about what’s next, I think about the potential developments in two categories or phases. I call them Generation 1 and Generation 2 AI agents.

Generation 1 is what we’re building right now, and we’re just at the start. Next up, the bots will not just be question and answer bots, but there’ll be a degree of familiarity and connection. They’ll introduce themselves. They’ll be friendly. They’ll also be personalized. They’ll look at your account information, know who you are, and not ask unnecessary questions. They’ll also be in all the channels that customers are in.

“While models are phenomenally sophisticated, they’re still in their infancy. We’ll look back at the current models and realize they’re still pretty basic”

Today, we chiefly do messenger; pretty soon, we’ll do email. Of course, we’ll do voice. So all the places where a consumer could ask a business a question, the bots will be.

Then, they’ll take action. You can divide up the category or the pie of customer service inquiries into informational, maybe conversational, maybe sales. But a big part of it is action. Not just, “Hey, do you ship to Canada?” But more like, “Hey, I accidentally shipped my shoes to Canada. Can you please change the delivery address?” That’s very high value. It’s much harder, but it’s what’s coming next.

When we launched Fin over a year ago, it was at like 25, 26% resolution rate. That’s where we think our primary competitors are at today when we run our benchmarks, and there’ll be a big race to continue to increase that. That will increase when it can do different types of questions, like taking action, but also when the models improve. Today, while models are phenomenally sophisticated, they’re still in their infancy. We’ll look back at the current models and realize they’re still pretty basic.

“The interesting question is not what’s coming. The question is, when is it coming?”

And so, through a combination of ingenuity and allowing Fin and these bots to speak on different channels and do different types of questions and the models getting better, we’re going to see these resolution rates climb increasingly higher, and we’ll have highly performant bots doing the majority of your customer service work.

The interesting question is actually not what’s coming. All the smart people in this space can imagine what’s coming. The question is, when is it coming? And as much as AI seems to have surprised us, actually, you can join the dots on where today’s AI has come from. I think it will take successive years for the key developments that I just mentioned to fall into place where the resolution rates are extremely high in doing the vast majority of the support volume.

Say hello to your new AI consultant

We talked about Generation 1. What’s Generation 2? Generation 1 is still stuck in conversations behind a little messenger screen. And that’s great. It’s really, really helpful. But Generation 2 bots will definitely break out of that medium. I spoke a little bit about voice. I think we’re going to see avatars. I think we’ll see video. I think we’ll be on calls with these bots.

They’ll certainly be proactive – they’ll come to you just like a clerk or a rep in a nice boutique store would. But they’ll go beyond the kind of “bot at your service” and start to be true consultants and experts. This is one of the truly transformative and shocking aspects of AI. It’s not just that it introduces efficiency, but actually it will be better than humans in many instances.

“You’re not just going to be speaking to a highly attentive rep. You’re going to get to work with the world’s best stylist”

We’ve got Waymos on the streets today. Those Waymo cars have five high-frequency LiDAR devices fitted on them with a level of awareness that no human can have. They’re not going to crash into people. People are even starting to choose Waymo instead of Uber because they don’t want to have to talk to someone, or maybe they feel safer. What’s really crazy about this AI stuff, again, is it’s not just efficiency, it’s actually superior and preferred in some instances.

Imagine you are buying a pair of shoes in this boutique store I imagined earlier. You’re not just going to be speaking to a highly attentive rep who can answer all the questions you have. You’re going to meet and get to work with the world’s best stylist. Not just a stylist, like, “Oh, they’ve got opinions about fashion.” No, the world’s best stylist.

And they will ask you questions. They’ll want to get to know you. They’ll want to see photos. They’ll give you interesting opinions and creative ideas. They might generate images to show you what these shoes look like with your favorite outfit. This will be an experience unparalleled compared to even the human experiences we enjoy today.

The Gen 2 agents will also be agentic in that they will have agency and they will go learn for themselves and become better. Today, with Generation 1 bots, you need to point them to information sources. You need to set them up, teach them how to do actions, etc. The Gen 2 bots will ask for a meeting on their first day. They’ll say, “Hey, Eoghan, I’m really glad that you chose to work with me. Can we please meet?” They’ll say, “Here’s how I like to work. Here’s how to get the most out of me. I’d love to get to know your preferences. Can you please get me on an invite to your All Hands meetings? I want to go to your exec team meetings. I want to go and do all the employee onboarding and training.”

“They will be whole customer service departments and organizations within themselves”

These agentic systems won’t just be a great customer service rep. They’ll be the best customer service leader in the industry. They will, after learning your business, say, “Well, here’s what Amazon did before and here’s their model. Modern smaller mom and pop stores on Shopify are doing it slightly differently. Here’s what I think will be great for your business.” They will be whole customer service departments and organizations within themselves.

You’ll also start to see people train customer service models as opposed to using third-party models. And all of that put together will transcend the Gen 1 experience that is a highly performant customer service chatbot to being a customer service, industry-best, expert system in a box. This is me using my imagination as best I possibly can, but the reality is we have no clue what that might look like.

The best of both worlds

We’re making a bet, and we’re either going to be right or wrong. Our bet is that humans are actually going to be around for quite some time, and our bet is that having humans and bots in the same system, tightly integrated, creates the best bots and the best human service and human experiences.

Today, you could buy Intercom, an AI-first customer service platform, tightly knit, integrated AI systems, AI Agents, and Copilot with all the human workflows and systems. Or you can buy a standalone helpdesk, like Zendesk, and then a separate bot, like either the thing that Zendesk acquired or a third party.

“Even as these bots get really good and we move from Gen 1 to Gen 2 and they can do all of these things, people will still want humans in the mix”

Of course, they’ll integrate them over time, but they won’t be seamless in the way that first-party systems and integrations we’ve built are. We’re pretty confident that, when they’re stitched together and the AI systems can access all the information generated by the humans and the humans can directly teach and interact with the bots, it will be superior. We think that’s just going to be better. And that’s why we think platforms are really, really important.

I’m pretty uncomfortable when I talk about replacing humans. I’m pretty uncomfortable when I talk about AI being superior to humans, simply because it could be misconstrued as my wanting to remove humans and humanity from our world, and the reality couldn’t be further from the case.

“The future is certainly not just humans, and I don’t think the future is just bots either. I think it’s going to be a beautiful mix of both”

I love humans and do not want to live myself in a world that’s devoid of human connection. What’s really, really interesting to think about is that even as these bots get really good and we move from Gen 1 to Gen 2 bots and they can do all of these things, people will still want humans in the mix.

I actually think that human experiences will become highly valued. I think that, in the very near future that we’re all about to live in, where you can buy great art, an outstanding meal, experience impeccable customer service, all by bots and AI, that we’ll be excited about and happy to pay more for art created by the hands of Mary, or a meal that’s the product of the passion of John, and customer service that comes with the empathy of Eoghan. Which again is another pitch for the AI-first platform.

The future is certainly not just humans, and I don’t think the future is just bots either. I think it’s going to be a beautiful mix of both.

Betting on the future

All of this has a few implications for businesses like Intercom in a big way and in a very big way. I think the first obvious implication is that companies like Intercom, whose revenue exists on selling seats for software for humans to do work, are under threat and will need to, if they want to exist in the future, cannibalize themselves by building software that does the work instead of those humans who used to buy the seats, much like Netflix famously pivoted and cannibalized their prior business that was sending DVDs by creating their streaming service.

“If we’re wrong, we’ll have built all the wrong shit and we’re in trouble. But if we’re right, we’ll at least survive, if not thrive”

People talk a lot about the incumbents having these great benefits, where they’ve got distribution and data and resources, whereas the little guys in AI don’t have a chance of catching up. One of the great disadvantages is that they’re set in their ways and they’re fearful and they have big bases of shareholders that don’t want to see them threaten their existing business.

Let’s see if it works for Intercom, but we’re certainly betting the farm on AI customer service. If we’re wrong, we’ll have built all the wrong shit and we’re in trouble. But if we’re right, we’ll at least survive, if not thrive.

“There will be a world soon enough where it’s not just the businesses that have agents working for them – the consumers will have agents working for them too”

The other very big implication for businesses like Intercom is that they may be disintermediated. This is not something you would tell a potential investor in Intercom. So I hope that either they don’t see this or they believe that this is a long way in the future. I certainly believe it’s pretty far in the future.

There will be a world soon enough where it’s not just the businesses that have agents working for them – but the consumers will have agents working for them too. And this whole Gen 1, Gen 2 narrative, where these bots elegantly assist and advise and connect and communicate with human consumers solving their every need, might start to become a little irrelevant when the humans have bots to do the work for them too.

What about in the future, when I want a new pair of shoes – for some reason, shoes is the example I’m using today – and I say to Siri, “Hey, can you show me a few examples of shoes like that pair I had last summer in Mexico that got wrecked on the beach? My budget’s this and I’d actually like it by Friday.”

“I can’t help but think to myself, what a time to be alive”

In the future – I don’t know when, but it’s not a million miles away, and it’s certainly in our lifetime – Siri will say, “No problem. What do you think of these? And that’ll be about the timeframe.”

And maybe that’s what shopping and interacting with digital systems and businesses looks like in the future. In that world, do you need Intercom? Do you need an agent? I don’t know. For my own comfort, I don’t like to think about that too much. Or maybe we’ll learn to adopt and adapt to that world too. Or maybe it’s the case that our business bots will talk to their consumer bots, and it’ll be bots talking to bots.

The reality is, what makes this so challenging and fun is that it’s incredibly difficult to predict what way this is all going to go.

You know, despite all these big existential questions and fears, despite how hard it is to constantly have to pivot and adapt and change in this world, I can’t help but think to myself, what a time to be alive. How cool is it that we get to contribute to this moment?

I don’t think in our lifetime, we’ll see the degree of change to society and humanity we’re about to experience in the next decade or two. I’d never want to be on the sidelines for that.

Obviously, I hope Intercom will win and do really well. I’m bullish. But the real winners will be the customers of these technologies. The incredible businesses run by hardworking, wonderful humans who can finally meet the expectations of their customers.

And the winners will be the consumers, all of us, who can finally get an answer to our damn question and get back to our lives.


Off Script CTA Horizontal



Source link

Leave a Reply

Your email address will not be published. Required fields are marked *