297: AI Beyond The Queue; with Robert Cabral

AI in customer support gets reduced to one tired idea: ship a chatbot and cut tickets. That mindset misses where the real leverage is. We sit down with Robert Cabral, Head of Customer Experience at Runway, to talk about “AI beyond the queue” and what it looks like when an AI-first company uses automation to scale service while protecting the customer relationship.

We get practical about the messy middle: migrating long-lived support reporting, pulling data from multiple sources, and turning scattered metrics into a clear weekly narrative that product and leadership can actually act on. Robert shares how AI helps teams find the story inside ticket data, platform usage, and escalations, and why reporting is only valuable when it drives a better customer experience, a better product roadmap, and a better day-to-day for the support team.

Then we move past talk and into action. We discuss chatbot foundations for product education, plus higher-impact automations that do real work, like refund handling with strict guardrails and human escalation when context is missing. We also explore how AI enables support teams to build internal tools and repeatable templates, even without traditional engineering backgrounds, and why that capability can become a career accelerator across CX, product, QA, and engineering.

We close on leadership: how to address the fear that AI will replace support roles, how to set customer-focused goals, and how to plan for different outcomes if AI resolves 90% of questions or falls short. If you care about customer support AI, CX automation, AI reporting, and building durable teams in a fast-changing landscape, you’ll want this one. Subscribe, share with a fellow support leader, and leave a review if it helps you rethink your AI roadmap.

Transcript

Welcome And Guest Introduction

Charlotte Ward 0:13

Hello and welcome to episode two hundred and ninety-seven of the Customer Support Leaders Podcast. I'm Charlotte Ward. Today, welcome Robert Cabral to talk about AI Beyond the Q. Today I'd like to welcome back after quite some time Robert Cabral. Robert, it's lovely to have you back on the podcast after what I think must be a couple of years, particularly my fault because I took most of last year off. So I take full blame on that. But it's lovely to see you again. How are you?

Robert Cabral 0:47

I'm doing great and great to be on again. Thanks for having me.

Charlotte Ward 0:51

No worries. I mean, always a pleasure. Um thank you so much for coming back. For the benefit of our new listeners, um, would you like to introduce yourself?

Robert Cabral 0:59

Sure. Uh my name is Robert. I'm based in New York. And currently I'm the head of customer experience at Runway, which is a generative AI platform for video and image. Doing a lot more than just that nowadays. Uh, but yeah, I've been in support and customer experience for over a decade, leading teams for for just as long. And yeah, really excited where the industry is is taking us nowadays. I'm excited to dive in.

Charlotte Ward 1:30

So, so really uh at the cutting edge of of AI as a product, never mind uh what we're gonna dive into in a second, which is how you're how you're using it inside the organization, right? But I love Rongway, I've made some fun videos out there. So uh yeah, um I'm I'm looking forward to this conversation because I I I know that you must be all in on AI. So um we we chatted a a bit about what we might talk about tonight, and particularly you drew on um you know your experiences of deploying AI, like actually like getting into the the um the task of like the the accomplishing this thing that we say that we call kind of yeah, using AI, deploying AI, uh, you know, um running support with the co-piloting with all of those things. Like I suppose, I suppose from my point of view, really open question to start with in terms of particularly support, but I know you lead a wider team there at Runway. Um, what does using

What AI Use Looks Like Daily

Charlotte Ward 2:30

AI look like to you day to day? And then I'd love to talk about how you got there.

Robert Cabral 2:36

Yeah, well, again, I work at an AI-first company. We our entire platform is built on AI, and we have a research team dedicated to building AI models as well. So it's very much encouraged for everyone to use AI on a day-to-day basis. And that goes from anything, you know, even outside of support, just any tasks that we have on a day-to-day basis that are repetitive, any reporting that we have in place where we have to draw from multiple sources, everyone's really enabled, including myself, to do what they need to do to make that happen. And yeah, so uh obviously looking at it from a customer support standpoint, when I first got at Runway, we actually didn't really leverage AI for customer interaction. And on the enterprise side, that was really because we, you know, wanted to be very protective of that relationship, those relationships and have that white glove support, which we still do actually. And yeah, when I was when I was brought on shortly after coming on and working on optimizations, that's when I was tasked with really looking at what what can we do as the business grows? What can we do to scale the business effectively without necessarily piling on headcount as has traditionally been the solution, you know, decades ago, for example.

Charlotte Ward 4:03

It it really was. And and I think it's the reflex for so many support organizations, even now, right? It's like we do all of our modeling, we do all of our forecasting, and I I, you know, I still I mean, I I think I'm j only just out of the habit of of forecasting headcount a year out, two years out from now.

Robert Cabral 4:24

Yeah, yeah. I still remember the spreadsheets of of doing that.

Charlotte Ward 4:28

Yeah, yeah, yeah. I I remember a good friend of mine, Craig Stoss, had a spreadsheet for uh a decade or more that just lived and breathed and grew like some kind of a monster. And I think every support leader out there has had this at some point, which is like what they what they kind of viewed as the source of all truth. It was a spreadsheet that never died.

Robert Cabral 4:50

Yeah, now uh now we're we're pulling those out with with Claude if we need to. Uh just going back to your earlier question.

Charlotte Ward 4:58

Yeah, 100%. And um, this is my experience right now. You know, I do have that spreadsheet still uh in my day job. Um, there are other bits and pieces that I mean I've I've refactored so much of what I do day to day already with AI, but I think that that monster is going to be the last thing that, you know, dies some sort of a gasping kind of death as I wrangle. How on earth I managed to configure that number that has been a number that I've been telling a narrative around for, you know, five plus years in this monolith, this monster of a of a spreadsheet. Um I, you know, we've all been there. Support is such a number-centric, such an operational part of the organization. I think to your point, reporting is one thing that in many ways is possibly the thing that makes me most nervous about swapping out for AI, but also is the thing that we should we should and should actually be relatively straightforward to swap out. Is is it I mean, you just touched on it there. Has it been a priority for you inside

Reporting As A Story Engine

Charlotte Ward 6:05

support to tackle reporting as a priority using AI?

Robert Cabral 6:11

Yeah, so I mean, just for context, and I think everyone is is kinda has always dealt with this. You have data from your ticketing system, you have data from your platform, what usage looks like, what how many monthly active users you have, you have uh escalation data that you have in place. So there's all of these data sources with very important information, and there's a story somewhere in there that's very important to communicate with both internally and to your team as well. You know, what what's the impact of what we're doing? What are our customers telling us? How should it apply to the product roadmap? And I think that connection so reporting itself is still gonna be the same thing whether we're using AI or not. You're answering the same questions, you're running the same types of investigations. The difference is being able to being empowered really as as even if you're not a technical person to bring all those sources together and really get to those stories maybe that you were missing or didn't have enough time to get to, and and drive for change. I I think you know reporting is important, but it's it's really a means to an end, and the and the end result should be a better customer experience, uh a better product, and also a better experience for for your team. So yeah, to answer your question, it is important, but everything that you can do with it, obviously, that's that's really what I'm looking to improve.

Charlotte Ward 7:51

Yeah, and and make repeatable to the point you made earlier. I I I would agree, uh, you know, the the word that I've been using fairly consistently for at least a couple years now, around all of my reporting is narrative, and and you know, as as you said, there are stories in that data that quite possibly only you can tell because you have the experience, but making that data, you know, less obfuscated or more uh, you know, making those reports more repeatable. I mean, who hasn't just, you know, for the sake of like one quarterly or one annual kind of opportunity, spent two days wrangling many, many systems just to get that one story we know is in there. But now the beauty of this kind of ease of repeatability is that you can tell that story on a monthly basis, or or more often, like you can it it can be a developing narrative rather than a painfully extracted narrative, right?

Robert Cabral 8:49

Yeah, exactly. Yeah, I used to do a quarterly report, uh, and I used to have a very light monthly report, and now I'm able to send out weekly updates using that that information. Uh it's still a little bit of time to put together, but not nearly as much time as you know, it took me before.

Charlotte Ward 9:09

Yeah, yeah.

Robert Cabral 9:09

That is driving to just quicker, a tighter feedback loop with product, quicker action, more urgency, and and I think just more transparency because a lot of times support can be a little bit of a silo. Everyone in sport knows what's going on, and nobody else seems to know what's going on. So this has not even talking about deploying agents, but just leveraging AI has helped us improve that internal communication.

Charlotte Ward 9:38

Yeah, yeah. Um I I I think for me the um the biggest hurdle to overcome is is just the the migration process for want of a better word. It's uh I'm I'm really on the brink now of of migrating some of my quite in-depth quarterly reporting over to something that will, you know, I mean, right now it take let's say it takes my quarterly report takes me half a day to do, roughly. Uh, you know, with some with some preamble, with a bit of a run-up of like collating a couple of figures from other places and what like maybe if I consolidated it half a day. To your point, I have monthly reporting, which I can do in about 45 minutes, pulling a few data points together, all of that. But like making the lift is an investment of time, isn't it? Because there is nuance, there is depth to that, particularly if you've been developing it for years. Sometimes, I mean, uh, like adding in other data sources, I can easily add in two or three more on top of the the common ones you you suggested. Uh I also have, you know, there's the workforce management data, so all of that shift mapping. I my team time tracks that's another data source. Um, and uh and we run a managed service, so there are metrics around like alert management and incident management. So it's a comp when you like the increasing complexity is something that like adds kind of a a slight um it adds extra height to that hurdle to to make the leap to uh fully automated AI reporting, I suppose. I am investing in that now, but it's not gonna be a you know, it's not gonna be a few hours work to make that happen inside an entirely new and to make it trustworthy, right? I can I could build something that looked like something but wouldn't necessarily be parallel to the figures I would draw and the the charts I would draw myself. Um how did you find that um migration? Did you find that you spent more time investing in reproducing

Migrating Legacy Reports With Focus

Charlotte Ward 11:40

exactly what you had, or did you look at it as an opportunity to do something different or somewhere in between?

Robert Cabral 11:48

Well, I I guess first of all, that this is not like a completed process. I I don't think it ever is. And the one thing to which you pretty much touched on is the more you're able to do, the more you want to do, and the the more complexity it can bring. So I think it's important to have a a real sense of focus instead of thinking about what can I do, what should I do right now that's gonna drive the biggest impact. And to answer your your latest question, it was really just building on top of what we already have and answering questions that we couldn't do so as easily. Um for example, we have we have in in our ticketing system ways of of tagging tickets and and understanding generally what people are writing in about. But frankly, that wasn't as effective. It's very manual. So that's one of the things that I really want to make sure we refine and it is an ongoing process. Understanding what are people writing in about. We have that anecdotal information, we can share it, but it doesn't carry as much weight when you don't have that data behind it. And that that really is what can for us, that's the first thing that we saw that could drive the the most change, had the most impact. So I guess now we we definitely have a lot more in place than we did before than I originally anticipated, and I think even the team is empowered to to build things out, and it's like, oh, I didn't even think about that. Um but that's and and I think it's kind of like uh how do you say it? Hurt hurting the cows? I'm not calling my team cows, but hurting cats.

Charlotte Ward 13:40

Cats I think of the animals you're after. Cows famously easy to herd.

Robert Cabral 13:45

I clearly did not grow up in a farm. So yes, hurting cats. Uh you know, you you have everyone that's super excited about getting all of this data done, getting these projects built out. And I think at some point you you really need to bring it all together and and bring that focus back in, especially as a leader. So yeah, the the migration process, I see it less as a migration process and more of building on what you already have. And eventually getting rid of legacy tools and reporting.

Charlotte Ward 14:20

Oh, please, I'm I'm so attached to that spreadsheet. It's scary. Yeah, I yeah, I think it is a uh, you know. And I think the other thing that I would just say is like not to underestimate the power of co-piloting with AI itself to make this kind of leap into you know building out more with AI. Um I I foresee the next few months to be very much an exercise in parallel running from my own experience, like and calibration, I suppose. I fully intend to run my existing reporting for a few months and experimenting to your point, picking the the big ticket items that I know will be either a huge time saver for me or be the ones that are I can predictably pull and like just hand over very easily. I think it's both ends of the the spectrum there, and in that parallel running, I'll do a lot of validation, you know, and a lot of a lot of pruning, I hope. Maybe I'll be I'll be willing to let some columns on that sheet go. In fact, I use that language today with Claude. Um, you know, maybe we can delete some columns next month. But um, but also, yeah, uh evol evolution, right? So building on what you have, building on what you have. Absolutely. When um when we get away from reporting, reporting is such a an art and a science, and weirdly repeatable, but also like iterative all the time. It's never finished to your point. That's why we all end up with a five-year-old spreadsheet with 32 ta you know tabs that we don't want to kill. Um but but when we think about beyond reporting, what what else have you done in terms of uh you know, picking up automations or deploying

Better Ticket Topics Without Manual Tags

Charlotte Ward 16:20

agents to do tasks or take some of the load? I mean, everyone immediately in terms of AI and support goes to AI support agents. You know, we're gonna deploy a chatbot, we're gonna have a co-pilot. Those are the really obvious ones. I assume you have both of those or one of those in in flight right now at runway. Is that a fair assumption?

Robert Cabral 16:42

Yeah, we we have a essentially a chat bot. And I think when we initially launched, we were really just thinking about it from we have all of this amazing knowledge on how to use the product. Let's leverage that knowledge to make sure our customers and and our prospects know what runway is, know how to use it effectively, you know, get get the best results and and not get frustrated with that learning curve. So that was the initial intention, and that worked really well. Once we really established that and realized that that was working, we started looking into okay, what are what are some of the actions that we're taking on a day-to-day basis that are important, yes, but very time consuming, and maybe not the best use of our team's time. And the first thing that came to that came up was uh refunding. We we have a consumer base and we have an enterprise base. So the consumer base, we we have a very clear refund policy and a very clear internal uh refund playbook. So we took the time to literally duplicate that with our agent to the point where we're absolutely sure it's only going to refund customers under these certain conditions, the same conditions that our team would. And when it doesn't have that context, escalate it to a human. So I don't as it as a customer, that's one of the best things that you can get, I think, when you are not happy with the service, you reach out and you ask for for that cancellation and instantly get that refund. So that has allowed our team to be a lot more strategic in the work they do. Uh and yeah, those are the types of use cases that that we're looking into.

Charlotte Ward 18:36

Yeah, yeah. The quick wins that aren't necessarily that are taking action beyond just talking to the customer. Yeah.

Robert Cabral 18:42

Exactly.

Charlotte Ward 18:43

Absolutely, yeah. Yeah. Um, have you had any successes inside the team that aren't customer-facing?

Robert Cabral 18:51

Successes within the team. I I think going back to the reporting, so our team is big on building out cloud projects. So the weekly report that I send out, that was actually automated not by me, but by someone on my team that was just proactively already doing that. So that is one thing. We also have team members just some some team members on on the enterprise side specifically get on customer calls, and they're showing a more complex part of our platform, which is workflows. And we've they they themselves have built out systems where hey, we're all talking about the same thing, repeating the same process, showing the same sort of workflow. Let's build out this template that guides customers through this process and allows them to build it out themselves. So we're leveraging the AI within the product, Claud within customer support. And again, a lot of times I'm not even the one that's that's driving it, which

Chatbot Foundations And Learning Curves

Robert Cabral 19:55

is great.

Charlotte Ward 19:55

So is that essentially product contributions from inside support? Essentially is what you're talking about. Yeah, I think that's become increasingly common. Um uh it's certainly happening day to day where I am, you know, uh product um innovation, product features being developed across the business. Yeah. Um and you know, I mean, I haven't gone quite so far personally as to contribute to product yet, but I feel like it's around the corner. But um, you know, having I mean, I've got three decades in support at this point under my belt. And uh when I left university, college, as you would call it, when I I did a computer science degree, I was coding um day-to-day in my career for the first 10 years of my career, and then I stopped because I went into leadership and I I really just essentially stopped writing code and like all of us, you know, started leading the team and leading the function. Um got to a point where it was just not part of my job. I I I and you know, air quotes, um, I, with the help of Claude, yeah, wrote some code last month, like and and developed an app and all of the, you know, like just just simple stuff like a few thousand line of Google app scripts to do something inside the team that was for the benefit of the team, you know. It was a Slack bot with some stuff behind it, and some of it was reporting, some of it was team-focused improvements, you know, but essentially engineering with a very small E, you know, but but stuff that was beyond my reach, uh, both in terms of time and up-to-date expertise, very few months before that, it's quite staggering. Um, the enablement, right?

Robert Cabral 21:40

Yeah. Yeah. No no one on my team has a traditional engineering background, and you know, we're building out Slack alerts through API integrations. We're with under certain conditions, building out those types of tools. So, yeah, to your point, AI is really closing that gap and helping us. I think across the board it's benefiting the engineering team, it's benefiting us. Although I I have heard some some complaints in general in the engineering world about vibe coding. So there's that. So I think it's just a balance. But definitely more more good uh than bad with the proper guardrails and guidelines.

Charlotte Ward 22:20

Yeah, and I think particularly on the innovation front, I think we I I you know I understand the engineering concerns, for want of a better word, around, for example, the quality of code and you know the uh the the how well it fits with you know how an organization generally delivers and things like that. But particularly at the innovation front, where it allows you to experiment very quickly and then get it to a point where you can hand it to an engineering team and say, go and battle test it, go and put it through, you know, the ringer in terms of our engineering function. But to take some of the heat off, like the the enormously quick innovation that's expected of any software house right now is

Automating Refunds With Guardrails

Charlotte Ward 23:05

um, and to to enable and empower the people who are using the software even more than the engineers, they're very likely to be your professional services teams, your support teams, even your customer success teams, your sales engineering, those kind of people. They're the people seeing customers day to day, seeing how customers are using products, seeing and listening to conversations firsthand where a customer says things like, I wish it would do this, or or even less like just, yeah, I'm trying to get to a point where, you know, even less obviously direct product feedback, but to be able to pick that up and just run with a fragment of an idea, it's staggering how quickly you can move and hand it over to the experts, I think.

Robert Cabral 23:46

Yeah, for sure. I and uh, you know, you you explained your career journey, and I I think you know, starting technical and going to the leadership level, picking up all those soft skills, and now we're seeing the opposite where everyone's starting off in in support. And you know, I have people on the team that end up going to engineering or product or QA, and that has always happened, but I think definitely now more than ever, when everyone's building out these tools and showing what they what they can build, but not not only what they can build, how they can think about the product has been a big difference from you know, even two, three years ago.

Charlotte Ward 24:30

Yeah, 100%, 100%. Um, I've got one final question for you, I think, which is you know, my my organization is AI native, your organization, I think you described as AI First, I think that essentially means the same thing, um, but uh broadly, right? Um, but um so so culturally and in terms of um just the people aspect of this, I think inside those kind of organizations, with the expectations that come along with working in that environment, with the tooling that's made available to our teams, um the opportunity that we hand to people when we work inside environments like that and give them that tooling. Um I think generally, or at least I like to hope that those individual contributors are engaged, are um future-thinking, forward-looking, um experimental um embracers of the technology made available to them. Um the the flip side of that, um, though, I think, is what we more often hear in the support community, which is the fear. The fear that AI is coming to take our jobs, the fear that as soon as we put a chatbot in place we'll need less support agents. Um I just want to get your read really on how other organizations that aren't perhaps quite yet ready to embrace AI first or AI native or anything like, like what do you um

Internal Tools And Support-Built Templates

Charlotte Ward 26:10

how do we talk to those support leaders? How should we frame this as an opportunity to them and and particularly enable them to have those conversations with their teams? That was a big question.

Robert Cabral 26:22

Yeah, that that's a very big question, actually. I don't know where to start, but let let me let me see. So I mean you see you see it in in the news, industry news all the time, even now, where an organization may decide, okay, we're going to go all in on AI support.

Charlotte Ward 26:46

At the same time, we're gonna do mass layoffs. Yeah.

Robert Cabral 26:50

And when that happens, you also follow along three months, six months later, and they're kind of backtracking and and they need to rehire all the people that that they let go. So I think that goes back to what are you actually trying to accomplish with AI? Are you just doing it because everyone else is doing it? Are you doing it because you want to save money? Are you doing it because you want to improve your customers' experience overall? The type of question or the type of goal that you have are really gonna going to impact how you roll out AI within your organization, how it's received internally, and I think ultimately how impactful it is, is it a positive rollout, is it a negative rollout? So that goal, in my opinion, needs to be very customer focused. You need to have a very customer focused goal, and yes, revenue is part of that at the end of the day. Sure. But there are different ways to go about revenue, and I think the best way to increase revenue is by again improving that that customer experience. Happy customers means more revenue. So yeah, there's a there's a lot going on in there, but I I think one starting with the goal to being transparent with your team and making them part of that process together. And even now, as you roll out AI, you don't really know what it's gonna look like post-launch a week after or two months after, three, three weeks, three months after. So yes, you should absolutely have a plan, and that plan should absolutely include everyone on your team to help lift that lift that up. Um and I think part

Leading Through AI Fear And Change

Robert Cabral 28:55

of that thinking should be, you know, what happens to the the rest of my team if this AI answers 90% of those questions, and it does so effectively. What happens if it answers 80% or not effectively? So almost having a path for each member on your team depending on how things go. For for context on my team, we we've always had a pretty small team, and I've rotated, I've had I've had a different team probably every three to six months since I've been here, and that's because everyone is is moving to different parts of the organization as we scale. As we adopt continue to adopt AI, everyone's focused on a different part of the business. And we allow them to lean into that, we encourage it, and that's ultimately better for the business when when some of our best hires are the people that have been in multiple teams because they're gaining all that context. Uh so anyway, long way of answering your question, but happy to dive into anything else if if helpful, or we can leave it there.

Charlotte Ward 30:05

Yeah, well well, it was a big question. I think I perhaps slightly unfairly landed that on you at the end of the at the end or approaching the end of the conversation. But you know, I I think it comes down in my mind to this in terms of that question. As leaders, generally, I think if you are opinionated enough, your team will hear you, and the way your team engages with this changing landscape is set by you. Um and I just think I I'm I am a simplistic person in so many ways, Robert, but I think you know it comes, it's pretty black and white as far as I can see at this point. You can see it as a kind you can play offense or you can play defense, you can see it see it as an opportunity or a threat, you can engage or you can be left behind, you know, and I think I think this is the future, like it or not. And so I am all in on using this to create those kind of opportunities that you talk about, you know, and I think like it's a whole other conversation, what they might be. And I've covered some of that on recent episodes, but like definitely want you to come back to talk more about your experiences and your team's experiences there. But but ultimately, you know, it's in my mind it's an opportunity, it's a set of opportunities, and I think you have to plan for that because otherwise um you are going to be operating in a landscape of fear, right? Rather than than uh rather than progress.

Robert Cabral 31:45

Exactly. And yeah, most things right now in this AI-driven world that differentiate organizations are the level of customer support that that they provide. So even I think the important thing to remember is even as you adopt these tools, and I personally encourage everyone to at least explore it, you should still see your team as a differentiator, as a competitive advantage.

Charlotte Ward 32:12

Yeah, yeah, I couldn't agree more. I couldn't agree more. Um Robert, thank you so much for coming along and sharing your experiences at Runway and uh ongoing experiences. You it's never ending to your point. It's never finished.

Support As Competitive Advantage And Close

Charlotte Ward 32:25

Uh and uh I think that's part of what makes it exciting for me as well. It's like there's just, I feel like constantly at the brink of discovery. So um thank you so much for sharing some of that with me today, with us today. Um, would you come back and dive into this a bit more at another time?

Robert Cabral 32:41

Of course, yeah. You'll you'll hear from me again.

Charlotte Ward 32:47

That's it for today. Go to customersupportleaders.com forward slash two nine seven for the show notes, and I'll see you next time.