Money 20/20 USA: Cresta CEO Ping Wu on when AI agents call your bank
- META
- CMCSA
- MAR
You can’t assume the caller on the other end of the line is human anymore. Ping Wu, CEO of Cresta, sits down with JD Durkin at Money 20/20 USA in Las Vegas to explain how personal AI agents like Meta’s Muse are starting to call businesses on behalf of customers, and what companies need to do about it. Wu shares how Cresta, the AI operating system for customer experience behind brands like United Airlines, Comcast and Marriott, built Sentry to detect whether a caller is a person or an AI agent, which agent it is and whether it has disclosed itself, so businesses can set clear rules for each. He also explains Cresta’s new Deputy Agent, which keeps working after you hang up by taking on the manual back office tasks that follow a call. And he makes the case that AI and human workers aren’t an either or choice, with people still handling sensitive conversations and AI learning from their expertise over time.
Ping Wu joins us here on the ground at Money 2020 in Las Vegas. He is the CEO of Cresta, which is actively helping to build AI technology that allows businesses to improve customer service. That’s welcome news to many people. Thanks for joining us. Good to see you.
Great to see you, JD.
Talk to me a bit more about the top priorities and the things you’re working on at Cresta in 2026, for anyone who might be a bit unfamiliar.
Right. So Cresta is a AI operating system for customer experience. We help some of the largest brand in the world like United Airlines, Comcast, um and Marriott to transform their business and is through, you know, with customer experience. Um and we provide a uh end-to-end AI platform to help them automate some of the interactions that they want to automate and and, you know, the customer does not want to wait for and augment their human workforce to make sure that they are more effective and more efficient and using AI and and also train them. As well as the third piece is really the analytics platform that we can see every interaction and surface the insights about their business and the process that they can improve on. So it’s end-to-end transformation of customer experience.
You’ve talked about the need to create a single AI operating system for the world of customer service. In your mind’s eye, what does that actually look like in practice?
Right. It means that there is no silos about the workforce, about your data, about your intelligence. Like, if I’m a customer, I call in or interacting with a business, I want the experience to be smooth. The customer really be seen as one customer from different touch points, from chat or voice. If I chat on website and then call in, I expect the business to know me and understand what get left off in chat and then continue that conversation on in the call. And also means that, um, the workforce, uh, the AI, now we have digital workforce as well as the human workforce. But sometimes I, as the customer, I speak to an AI and also I speak to the human by escalation. I want that experience to be a smooth experience as well. So, I know the the context not get lost, and then it’s a just one continuous customer journey. So that’s why we believe that, you know, in order to really transform, truly transform customer experience, we need to meet customer where they are and support all the touch points with one AI platform that automating interactions, but also augmenting human workforce.
I want to ask you about Sentry, a recent product, I know you and your team at Cresta, I’m sure very proudly have introduced. It helps businesses better identify AI agents who are calling on behalf of customers. How does that actually work? And why is that so needed in the ecosystem today?
Right. So today, um you know, personal AI agent calling on top on behalf of consumers is the top of mind for most of the businesses. And we see the popularity of the personal AI agent like Muse or Open AI launch another one called Dot and then Instinct, there’s a is a separate startup that uh build those consumer AI agents. Um business now really get interaction, uh not only just from their human customer, but also from uh those AI. Um then, you know, it’s very important for them to first detect whether the caller, you cannot assume the caller is always human anymore. You, you know, there may be another computer somewhere else, right? And then Sentry is the detection model, uh to be really the first layer to understand whether it’s a human or AI. If it’s a AI agent, whether it discloses herself or not, with which uh, which AI agent is it? Is it Muse or is it Instinct or is it some rogue AI agent that does not even disclose themselves. And then that will allow businesses to set up policies on what human can do, what AI can do, so that they can really control that experience.
How about deputy agent? I think the idea here being I’m the customer, I hang up the phone, but this thing is still often working. What does it actually do for me as the end consumer?
Right. So, um, JD, a lot of the business processes, uh in enterprise do not start uh until the call finished, right? So when if you call in the call center or you do something on a website, you know, it’s not like the everything is done when you hand up the phone. You know, in fact, a lot of things actually get started. If I call in to change my PCP, um the call center agent will flag a ticket and then there’s a entire back office team that made that possible end to end. Or if I call for a complaint, a bank, there’s an entire back office team doing those bank um doing those complaints, um, pro process. And a lot of these process are very manual. It require you to go through multiple screens and, you know, multiple teams. And we believe that, you know, the, the impact AI agent can make for customer experience should not just be customer facing. And we think about it’s on the phone and automate the, like chatbot, but also take the experience end to end, really take away those business processes that downstream from call centers and then also automate those. And so that you can get those, you know, operation more efficient and also potentially customer experience can have that more consistent context from the call centers all the way to the back office to get things done for them.
People will hear this conversation and they will wonder, understandably, what is the role of human employees in a, in a world where AI powered technology is doing so much of the customer service work that, you know, maybe it didn’t do a few years ago?
Right. So we believe that, you know, AI and the humans are, you know, they’re both needed. It’s not either or choice. And then just, you know, they’re different type of need when the customer interacting with business. Sometime, for example, they may actually do not want to even interact with business. Like JD, if something is working as you expect that, you probably do not want to even call in. So that’s, you know, our view is business need to understand all the interaction, get the insights and then fix the problem if there is a, um, there is a thing to fix so that the call can eliminate. And but there are also calls that consumers they do want, do not want to wait. It can be very highly transactional. They just want to get answered and move on or get things done. For those, AI agent should absolutely automate. But for a lot of interactions today, uh, majority of the interaction, customer experience, still human to human. And a human sometimes really prefer another human for some more sensitive emotion topics. And for that, we allow the consumer, give them the choice to speak to a human. So we believe that’s not either or. In in fact, that’s why, you know, we’re talking about the unified AI platform, the AI operating system, they need to do both. Both, you know, the interacting, automating with AI, customer facing, but also at the same time augmenting a human workforce so that you know, overall gets better. And then there’s a one last thing is the feedback loop. I think that is very important, that we do not believe that building AI automation in silos. AI should really learn from humans because your business change and evolve and over time your call center evolve as well like but the best expert are the humans. And then that’s how, you know, really need to feed on the human expertise and then make the AI better. So overall, we see that, we see this is really a growing pie. It’s not a zero sum. There’s a lot more things AI can do in business, especially in customer experience, in my view without even hurting jobs.
Finally, let’s take a look to the future. Pretend it’s five years from now. It’s Money 2020 2031, you and I are on the ground here in Vegas. What does the world of customer service look like in that conversation?
I think that first of all, the overall customer experience will be back potentially just by one AI. That’s almost like a central nerve system that understand all the interactions that improve certain aspects in an organization by itself. You know, it’s improve the processes so that they can discover anything broken so that people do not have to call in or give the feedback to the product team so they can improve. And then it will also observe all the, you know, the automation opportunity that AI agent can take by themselves. But also at the same time, you will really oversee all the human conversation as well. We’re doing that already, but I think five years from now, we absolutely just, all these are not separate AI system. It’s a one AI system that oversee all these interaction, improve every aspect of the business, improving the human operations, improving the customer facing AI operations, improving the business product by itself so that the customer experience can get better. So that’s, you know, what we feel. And then maybe one last thing is we do think that five years from now, more and more of those calls, will, especially those more maintained ones will be done by AI. So the customer may, you know, not even calling themselves. They would delegate the ones that they do not want to have those interaction to AI. Um so, yeah, it will be a very, very interesting five years, probably the most, um, change we will see in customer experience in every five year period will be next five years.
Ping Wu of Cresta, outstanding interview. Thanks for the context. Thanks for the conversation and the time. Really nice to see you.
Thank you so much. Thank you. Thank you, JD.
Thank you.
