Author transcript
Sean Nolan: Hi Victor, thank you so much for your time today and for joining me on CX Today. To start things off, can you tell us about iVisa and your role as Director of Customer Experience?
Victor Gimenez: Of course. Thanks first of all for having me here, Sean. It's great to be here at CX Today. So iVisa's goal is to make sure that users focus on their trip instead of the paperwork. What that means is, every time a user has a trip, they have to get their travel documents, or visa, their ETAs, whatever it is. And it's always a big hassle before they can even start thinking about their trip. iVisa's whole goal is to make this as easy as possible so users can get from wanting to travel somewhere to traveling there as easily as possible, and not think about all the travel documents needed.
My role in all that, and in CX, is making sure that actually happens and making sure the experience is very positive, and working with the other teams to improve the experience and make it as frictionless as possible. This is especially a big pressure for CX or for customer service, because a wrong answer can lead to a customer stranded at the border in another country. So there's a lot of pressure for us to make sure we get it right, and if something happens, that we make it right afterwards.
As Director of Customer Experience, I lead the customer service team as well as the AI assistant, and also the overall experience initiatives we have across the board. So leading NPS, CES, and canceled requests or canceled orders, to make sure that everybody at iVisa is thinking about the experience. Even if it's a growth initiative or an operations efficiency initiative, everything we're doing has a customer at the center. That's where we come in. We give the data, work with them, and actually find a solution, and we're the owners of that solution itself at iVisa. My goal is to make sure that we are not only the voice of the customer and a reporting team, but actually responsible for the solution and for improving the experience at iVisa.
Sean Nolan: Thank you so much. Sounds like really important work. You mentioned AI, so I wanted to start by asking, where do you see AI adding the most value to the CX teams at iVisa?
Victor Gimenez: This is great, because AI has been so progressive and we've been learning and growing along the way. I think a year ago I wouldn't have been able to answer that question as precisely as now that we've learned so much. I would say that it helps us understand our customers better and also serve them better. The understanding part is more about CX than the customer service side. CX as a whole, really understanding the opportunities that we might have to improve.
As any CX leader knows, it's easy to spend a lot of time looking at data. There's so much data, especially in tech companies. If there's something that we have, it's data, and at iVisa I think we have an amazing data team, so everything is available. It's just really hard to crunch those numbers and really understand what those numbers mean, what the surveys mean, what the comments add up to, and get to conclusions that explain what's going on and what we need to improve.
Before, we sat with the team on Monday and started doing the whole analysis for the week before. On Wednesday we got together and asked, okay, where are the deep dives? What do we need to look at? And if we were lucky, by Friday we really understood exactly what went wrong. But then Monday came again, the numbers could have changed, and we found new opportunities. Sometimes it could become this loop of just getting data, data, and the backlog of initiatives started building up.
Now with AI, we have a scheduled report that looks at all the variables, even more than we used to look at before. It does this before we even wake up. By 8 a.m., where we're connected, the report has already been sent to all of us. So we can start from step five, do the deep dives, and ask the correct questions. The main thing is that the team focuses much more on solutions than on just running around seeing what happened last week. At least for me in CX, it's become that curse where you want to know, you really want to get to the bottom of it, and you really want to solve the correct issue. So you want to make sure you understand it, but at the same time you need to really focus on solving these issues.
On the serving side, our AI assistant has been the best example of this. It's something we've been building for a year and a half. We've built it internally, on top of LLMs, but we don't use a third-party vendor. So it's really been a journey of understanding how to build this and get this right. We have a team that manages it constantly and makes sure that our customers get the best answers from our AI assistant.
Sean Nolan: Thanks so much. To touch on what you mentioned about the serving side, I understand it's really important for iVisa, because you cater to international audiences. What's been your experience with AI assistants and translation, and how is your team thinking about keeping these AI assistants accurate and transparent across different languages?
Victor Gimenez: That's a great question, because it's, I wouldn't say a problem, but a great situation to solve, and to really find the best mix of how to get to the right answer and the right solution. Obviously, our main language is English, in the platform and in the team itself. So most of what we build with our AI assistant is done initially in English. Then comes the part of how this translates to all the different languages we offer. The chatbot, or AI assistant, I should say, handles even more languages than the 14 we have on our site. We try to answer in any language the customer writes to us. So we write the prompt in English first, and then we test it in the different languages, the 14 we have.
But there's always the possibility that in a real case, the translated version doesn't resonate or have the same effect as we wanted in English. So we have a dedicated CX AI team. Inside the CX structure, we have an AI team that works hand in hand with a tech team. We do daily QAs to figure this out and find where we might be affecting the user. We do regular, randomized QA, and we also do QA for negative CSAT for our chatbot, to find those specific pain points. Until now, we haven't really found a big opportunity in what the bot says. It's usually very accurately translated from what we have in English.
What we have seen, and the big challenge we've mostly had, is with the bot understanding triggers from what the customer says. Sometimes we want to build specific triggers for the AI assistant to do something specific. In English, we select the specific words: this might mean this, even though the customer didn't technically say it, when he says this, he's probably referring to this, so take that into account. When we build those, the whole team has to be on top of the launch and see if anything is interpreted a little differently, because we have seen that in other languages it doesn't catch that trigger. So it takes a little bit longer for the AI assistant to realize, okay, that's what the customer means.
The best way to solve this is mostly having our QA team be on top of every launch, before and after. In testing, we make sure we test as much as we can, which we always try to do, and then after we launch, we keep the QA going. That's true not only for launches. We always have a concurrent QA team that checks conversations and tries to fix anything we find, because those things do happen.
On transparency, it's always the question of how the bot should introduce itself. I think the final, clear answer is that the bot should definitely say initially, "I am a bot, you are talking to an AI assistant." I think that should be very clear. That's what we're doing with all of our conversations in every language. We have that hard-coded in our bot to always say at the start of every conversation.
Sean Nolan: Absolutely, thank you so much. It's been really fascinating to learn more about the CX foundations and the work that you're doing at iVisa. With all this in mind, I wanted to get your perspective on how you're finding the balance between AI agents and human agents in the CX journey. How do you maximize the value of both sides to get the best customer experience at the end of the day?
Victor Gimenez: That's the big question when we talk about AI and customer service. The best way to look at it, at least initially, is that we should always start off with what the AI assistant can do better than a human, and what a human can do better. Obviously, both can change over time, and I'll mention that a little later. But understanding that first makes sure that you have an AI assistant that is actually there to improve the experience, and not just there to reduce the workload, which is how it can also be thought of. If there's a team that shouldn't have that mindset, it's definitely CX. It should always be experience first.
In our case, our AI assistant started off focused mainly on informational requests. What's the status of my order? When's my order going to get here? Am I going to be able to travel? I got my visa, what does this mean? Some visas and documents are very hard to explain, and the AI assistant can directly answer those conversations. It can do it a lot faster, and we can make sure it's fully complete and not missing anything in every one of the answers. So initially we start off with that, and then the humans can do the more complex, case-by-case work. I'm at the airport and I need some urgent help. Or I'm uploading this document because I have this situation specifically in my country where I can't upload an overall passport, or whatever it is. They need a deeper-dive answer to get a solution. Having that split makes sure the experience starts off with a good baseline, where we're doing the best in both cases.
Then the way you grow is you try to teach the bot what the humans do best and start doing that little by little. It's sometimes tempting, I can't say it's not, to just turn on the bot and handle everything. But you realize that the experience just goes down, CSAT goes down, and the customers are just not happy with just having a bot answer everything, unless it could do it equally or better than a human.
As I mentioned before, sometimes that mix can change, and that's happened at iVisa, which I think is very interesting and very healthy. We had the AI assistant handle new conversations. It was doing it better than the human team because of how we trained it. Then the human team came up with a new process, worked on a new solution and a way to improve the experience, and started doing it better than the AI assistant. So we reverted it back. Now the plan is to ask, in the future, can we get the AI assistant to do it as well as the humans are now doing it? When we have that relationship work for a better experience, it's super healthy, and it's what we strive to do. Hopefully that keeps happening a lot.
Sean Nolan: Yeah, absolutely. It's really interesting what you mentioned at the end, about being tempted to use the AI for a greater number of services. But in the end, if you're seeing things from an experience-first perspective, that's when you see the value of humans coming in. That speaks to the value that both sides can provide in the CX journey. For people watching this who also work in the CX sector, I wanted to ask, what's one key piece of advice you have for CX leaders?
Victor Gimenez: In terms of AI, and especially the AI assistant, my biggest learning has been that it's not plug and play. It's not, okay, let's just get someone to build the bot, it's going to be built, and then we just let it run. As CX leaders, we're used to putting a lot of effort, resources, team structure, and strategy into building our human teams, to make sure they offer the best experiences. We have QA, we have process teams, we have training. We have all these support structures for the human team to improve day to day. We have daily QAs, one-on-ones, coaching, an endless amount of resources for the humans. Then we go to the AI assistant and we cut those resources by 99.9% and just try to turn it on.
I think the learning is to think about it as, maybe, a leaner human team, but don't do anything initially differently from what you would do for a human team. Start off thinking like that, and then say, no, I don't have to do this because it's not a human team. Maybe I don't have to do coaching, because it's not a human team. But start off with the same things and start cutting what doesn't make any sense at all. I think people are going to realize that most of it is going to stay there. The process team is not a process team, it's actually the prompt team, but it's basically the same thing.
QA is probably even more important, because an error in the prompt is going to be repeated a thousand times exactly how you wrote it. So even more QA than you had before, and improving it day to day, week by week. Working very closely with tech is something very, very important. Just as you would work very closely with HR on the human team, on the AI you need to work as closely, or even closer, with the tech team. It's the equivalent of making sure everything works out and you have the team, so to speak, that you wish for.
If you put in enough resources and you realize it's an ongoing team and you need to improve it day to day, you're really going to get to where you want to go. But don't expect it to be plug and play, or an easy way to have some conversations and not reach my team. Have it as a core part of the strategy.
Sean Nolan: Yeah, absolutely. That's really practical advice for CX leaders as they adopt a new AI tool: to have the same strategies and foundations in place as they would for a human team, like QA and looking over the processes. I think that's a really good way to get the AI functioning and delivering the services that you want it to from the get-go. And that constant evolution process is really important. Thank you so much. My final question is a forward-looking one. What are your predictions for CX in 2027? Are there any opportunities on the horizon that you're excited to explore?
Victor Gimenez: Yes, one hundred percent. The main thing with AI is that we're all on that learning curve of using it for the best or biggest benefits we can. For CX, the possibilities are definitely endless. One thing I'm really excited about, and I think it's even more than the tools for AI, is how we use it, how we incorporate it and understand the power of it. For me, it's more about how we use the technologies we have now, and AI, to start reducing contacts altogether.
I think most CX leaders have contact rate as one of their main KPIs. It's always been there, and we always work with product and try to get that connection to fix these issues. But it's still always very after the fact. With AI, my hope is that we can use it to really prevent these issues from happening, case by case, and AI can really do that. One example for iVisa would be having the form checked while you're filling it out. So not fill it out and then get a response, but have real-time checks with AI. This user is having this issue, or this user didn't enter the correct input, and we could call them, or even the AI could just give a suggestion before the issue even happens. I think that's what we need to strive for.
I think it's the best time to think about the contact that never happened, and to start talking, even more than contact rates, about contacts prevented, which I think was impossible before. I would have loved to have that metric with all the work we've been doing, but you don't even know where to start and how to measure it. How do you know exactly if you prevented it or not? With AI, we're now going to be able to solve these issues before they happen, track them, understand them, and keep adding to them continuously, to really become the role of CX and make sure that users don't even have the issues to begin with.
Sean Nolan: Thank you so much. What you said there, the contact that never happened, I think that's a really groundbreaking metric. It changes a lot about the traditional things that CX leaders have been keeping an eye on. This is a new metric to consider, and it's a really exciting time. I appreciate your perspective on that. Thank you so much, Victor, for your time today. It's been great.