AI, Customer Experience, and the Power of Starting With the Problem

In this episode of The CX Equation, hosts Chantelle Casey and Mark Clydesdale sit down with Zeke Wu, Global CRM Program Manager at Nissan, and Aswin Peter, Senior Product Lead at one of the UK’s energy giants, to talk about what it really takes to use AI to improve customer experience.

Both bring hands-on experience from large organisations applying AI within customer initiatives, on a global scale. Drawing on lessons from the automotive, energy and other sectors, the conversation quickly moves beyond the usual AI hype and into the practical realities of what actually works.

One theme is key: Instead of starting with the question “What can we use AI for?”, Zeke and Aswin argue that teams should begin somewhere much simpler: What could we be doing better? Only once the problem is clear does it make sense to ask whether AI is the right solution.

In several cases they discuss, the biggest improvements didn’t actually involve AI at all. They came from fixing fundamentals – simplifying journeys, improving processes, connecting data that wasn’t being used effectively – i the process.

This episode covers:

• Why problem-first thinking beats AI-first strategies

• How trust and transparency influence AI adoption

• The practical ways AI is already reshaping customer service

• Why global AI frameworks still need regional flexibility

• How organisations can start creating value without perfect data

• Why competitive advantage ultimately comes from better decisions, not just better tools

The discussion also explores how AI is beginning to reshape customer discovery and brand visibility, particularly as search, recommendations, and digital assistants become more intelligent.

You can listen to it, or watch it, on all major podcasting channels. Here are links to the favourites:

Spotify
YouTube
Apple Podcasts

00:09 – Intro
Welcome to the CX Equation, a podcast by TapCXM. We share actionable insights and real world case studies to equip you with the tools you need to drive loyalty, engagement, and sustainable growth. We’re your hosts, Chantelle Casey and Mark Clydesdale

00:25 – Mark
On today’s special episode of the CX Equation, we’re delighted to have two guests with us. We welcome Zeke Wu, Global CRM Programme Manager at Nissan Motor Corporation, and Aswin Peter, Senior Product Lead for Customer Experience and Product Strategy at British Gas.

00:40 – Chantelle
Zeke is a senior leader specialised in AI-powered sales and customer strategy with extensive experience across automotive, aviation, and consulting. He currently works at Nissan where he’s leading AI-driven CRM programmes in global customer strategy. Zeke’s work blends data science, predictive models, and customer insights to enhance both customer acquisition and long-term value.

01:02 – Mark
Aswin brings a wealth of experience in product management and customer experience across the energy and utilities industry. Currently at British Gas, he’s leading the open banking strategy and spearheading initiatives at improving energy services for residential customers.

01:15 – Chantelle
Aswin is also a member of the Perplexity AI Business Fellowship, so we’re about to have a really interesting AI chat with Zeke and Aswin. We’re going to explore how AI and data-driven insights are powering innovation and customer experiences, the challenges of implementing AI-powered strategies, and the tangible outcomes these technologies deliver. So yeah, let’s get into it. Welcome to the podcast, Zeke and Aswin.

01:37 – Zeke
Thank you very much for inviting us.

01:39 – Aswin
Thank you.

01:40 – Chantelle
More than welcome. We’re happy to have you here. So you’ve both worked across multiple industries. Zeke, you’ve worked in aviation, consulting, and automotive, and Aswin from manufacturing to energy. Tell us a little bit more about how you got to where you are and what first sparked your interest in using AI and data to transform the customer experience. I’ll start with you first, Zeke.

01:58 – Zeke
For me personally, I started my career on the technology side. It has always been about using technology to improve the customer experience, sales conversions, or customer service satisfaction, and so on. Over the years, as I shifted location from China to Finland to the UK, I also shifted from company to company and industry to industry. One of the things I observed is that technology has always been an enabler, and now it’s an accelerator. Previously, it was called digital transformation, or even before that, it was called IT, then digital transformation, and now AI. Everything is speeding up more and more, and we’re actually seeing trends increasing the pace towards a more human-centric approach assisted by technology. AI is the latest trend, but it’s definitely not the last one, and that naturally feeds into my daily work every day.

02:55 – Chantelle
How about you, Aswin? How did you get to where you are?

02:58 – Aswin
I started with technology consulting. The good thing about consulting is that you get to work with different customers, different clients, different industries. You start seeing patterns in the problems they’re facing, and that excited me — recognising that there are common patterns across industries and that there is a way to solve them. As Zeke said, technology is always an enabler. It’s not the end goal. We always need to start from the problem — what is the problem first, and then approach it backwards. That’s been my motto. I’ve always looked at what customer problems are, what problem we are solving, and whether the problem is worth solving. That’s how it started, and eventually I moved predominantly into the energy and utilities industry, where I’ve been for close to a decade. I’ve worked with multiple customers within energy and utilities as well. The same things hold true there — there is a particular pattern in what customers’ pain points are and what they’re trying to solve. AI, like any other technology, is an enabler, but you should never go in with AI first. It should always be problem first, and that’s how I’ve looked at it.

04:05 – Aswin
AI has always been there. It’s not something that just came into the picture now. Generative AI and agentic AI have caused a lot of buzz, but AI has always been there. Predictive analytics have always been there. We have always been using AI. But you can now see AI and play with AI. Earlier, it used to be like a black box where people would run algorithms and then see the output. That’s the difference. What has changed is how AI can impact the customer experience directly, rather than a back-end model that runs and gives you some output. I think what’s exciting is how it’s heading forward and how it will evolve the customer experience going forward.

04:47 – Mark
I think that takes us straight into our first question, which is how is AI actually impacting customer experiences in your industries right now? Aswin, can you tell us a little bit about what you’re seeing AI doing in the energy sector?

05:02 – Aswin
Absolutely. The way I see the maximum impact is across three practical areas. The first area is essentially self-service that actually resolves issues, not deflects them. Customers today want their issues resolved as soon as possible. They don’t care what you employ or what AI models you use in the back end — they want the issues to be resolved. Today, if they want to get an issue resolved, they don’t want to pick up a phone, hold on for forty-five minutes, and then have an agent try to investigate the issue and deflect them. That’s not the experience they want. They want things done through self-service, and I think AI is playing a great part in enabling that. Customers don’t want just FAQs. They want something they can interact with that can actually understand their issue, not a generic FAQ. That’s the first area. The second area is agent augmentation. This is a very crucial area. Today, agents struggle because when a customer calls, they don’t have background on what the customer is calling about, so they have to start from scratch. Imagine a world where AI can summarise the customer’s situation in front of the agent so they can say right away, “I see this is your issue — your payment didn’t go through last time.” It’s tailored and personalised to the customer. I think that’s a game changer, and it really enables both customer experience and helps the agent resolve issues and have a meaningful conversation. The last area, the way I look at it, is proactive service. Customers today do not want to wait. They want us to understand them and reach out to them with something tailored to their needs, rather than just bombarding them with anything that isn’t relevant. So those are the three main areas. Holistically, the way I look at it is that customer journeys are transforming into intelligent journeys. What I mean by that is that experimentation should be adopted across how the customer journey progresses. Earlier, we had a predefined customer journey — customer clicks on this, moves to the next. With the introduction of agentic AI, those journeys may look very different. I think we need to understand how that transformation is going to happen in the customer experience world, and I think that’s going to be pretty crucial and exciting — how things are shifting.

07:21 – Mark
Similar to what you’re experiencing, Zeke, or is automotive completely different?

07:26 – Zeke
It’s not completely different. It’s actually quite similar. Aswin summarised very well that there are effectively three types of AI. Traditional AI and machine learning — that’s been happening in the industry for many years. Then generative AI, and then agentic AI running in the background, offering insights and helping companies make decisions, improving efficiencies — all of which reflects back to the customer experience the company can offer. It’s not straightforward. On the other hand, the hype around generative AI and agentic AI is actually helping traditional AI gain more strategic investment, and that’s accelerating how AI is helping to improve customer experience in this phase. For generative AI and agentic AI itself, we have been introducing some tools internally to gather customer feedback and understand customer intentions, and that information can be used for decision-making — to find bottlenecks where customer service isn’t good enough or where customers are encountering difficulties in a specific journey. Those things still reflect back to the customer journey and result in better service and greater understanding of how the company can help them. There are also external challenges posed by generative AI and AI search specifically. Consumers are no longer relying on company websites for trustworthy information. They go to Google, to Perplexity, or to ChatGPT and ask, “What SUV suits me and my family with two kids, two dogs, and we want to go camping?” It used to be that customers would search on Google and find out SUV capacity, size, and fuel efficiency. Nowadays, they rely on Google AI Overviews or ChatGPT to give them a recommendation. That’s a growing area we’re now investigating — how should we be present in those early phases? And sometimes those customer questions come at a later phase too, like how to fix something. These are all changes in the consumer journey and consumer preferences that companies — not only automotive but across different industries — should be aware of. Start monitoring it and create your own influence to make sure you are present and performing well in these new, challenging situations.

10:11 – Mark
How do you do that? Because ChatGPT — and I’m sure any of the different AI large language models — would say they’re always going to answer the customer’s question first. So how do you, as Nissan, make sure that ChatGPT, when it’s looking at what the best SUV is, knows that you’re a valid option for what that customer needs?

10:33 – Zeke
Basically, it boils down to how ChatGPT selects and summarises the best answer. They use a lot of different techniques and algorithms to determine which information is more trustworthy. For example, Wikipedia is most of the time very trustworthy — sometimes not, as we know. And if content is referenced across different websites and channels multiple times, it’s considered more trustworthy. So they use that information to summarise and feed the answer. For corporations, companies, and industries, it’s about making sure your content is consistent across channels and referenced as much as possible. That becomes a word-of-mouth effect on the internet and can help you. On the other hand, it’s also about what questions customers are asking. They might be asking, “Recommend me a car,” or they might be asking, “Compare a Nissan SUV to a Toyota.” Those different questions trigger different search algorithms behind it, and you need to know what keywords or key questions and intents you want to be present for. Otherwise, it becomes a limitless space you’re competing in, and that’s not possible.

11:55 – Chantelle
Just talking about real-world implementation then — we’ve heard a lot of buzz about AI, but I think only recently, over the last six months to a year, have we really started talking about actually solving problems rather than just using AI for the sake of it. Can you walk us through any use cases where you’ve deployed AI and it’s actually solved a customer problem?

12:16 – Zeke
For AI use cases, a lot of industries are quite advanced and already have customer-facing implementations. For us, it’s still mostly in the back office. There are two cases. One is that we’re actually using generative AI to analyse customer feedback from multiple channels. The voice of the customer, through our own websites or through third-party channels, is collected and analysed, and that feeds back into different areas of the company. The use case itself is very significant. Some of our most important regions are using that information to inform their daily business operations — to improve sales, customer service, or dealership performance. The other use case is still internal and not directly customer-facing, but it does impact the customer in the end — and that’s knowledge management. We have lots of data internally, and it’s buried in various places, and not everyone can access it easily. One of the challenges we face daily is: can we get this or can we find that? It becomes a huge rabbit hole as you dig through the organisational network. One of the things we’re trying now is feeding that data into AI. There are security and governance concerns, of course, but once that’s ready, the AI will essentially be a super-agent that can help us find all the insights we need.

13:41 – Chantelle
Aswin, is there anything you can walk us through, and any challenges you hit during implementation as well?

13:47 – Aswin
Absolutely. When it comes to the energy and utilities industry, we are a highly regulated sector. These new AI models are all non-deterministic systems. What I mean by that is that you have an input, but you cannot guarantee an output. There can be a deviation every time. That means it’s not very difficult to build a proof of concept for an AI model or a feature, but the problem comes when you try to scale it without a proper evaluation framework in place to ensure what you’re getting is accurate and performing as expected. Those two areas are very crucial. Especially in energy and utilities, where we are highly regulated, if something goes wrong, the repercussions are too heavy to handle. So we take a very cautious approach towards AI. However, having said that, AI can actually elevate customer experience in multiple areas. I look at it as an enabler. It’s not a replacement for human thinking or decision-making — it makes things much more streamlined and helps customers overall. A couple of things come to mind. Earlier, when I worked at a manufacturing firm providing products to the forecourt — essentially petrol stations — all the IoT devices were connected. Every time you drive to a petrol station, you see a price displayed. Traditionally, what most of them do is have a store owner who maintains an Excel sheet with all the prices and essentially takes a bet — the petrol station nearby is charging this, so we should probably reduce ours slightly. What we actually did was create a platform for price management, which intakes all the prices and makes a decision. That is an AI and machine learning model running behind it. But it doesn’t stop there. The store owners are not very tech-savvy, so they want to converse with the system — for example, if I change this price by this much, how much will my margin be affected? They can converse with it much like ChatGPT, but in a controlled fashion, and make decisions much more quickly rather than working with Excel. That was a real game changer, because traditionally the forecourt and midstream industry is considered a laggard — not as advanced as, say, retail. For them, this was transformative and really made them more efficient in the long run. Turning to the energy and utilities industry specifically, a couple of areas where I see massive improvement happening are around contact resolution. Today, customers contact us through chat, but they don’t get answers. If they say, “Why is my bill high?”, all they get is a link to a list of possible reasons. There’s no real conversation. They feel like they’re talking to a bot that doesn’t understand their context or why they’re contacting. If they persist, they get deflected to a human agent who has no idea what the issue was, so they start from scratch again, and the customer gets frustrated. AI can actually help us move from a static FAQ approach to understanding intent better and answering the customer in a human way with quality answers and resolution — helping them resolve the issue there and then, and reducing the contact rate. That’s a huge area, especially in utilities where you’re dealing with millions of customers. Even if 10% of those customers start calling, we can’t serve them all efficiently. That’s where AI can help — not just for chatbots, but also from the agent perspective. If an agent is speaking to a customer, they should have all the relevant information right in front of them. If the customer has already raised a complaint, that should all be summarised. They shouldn’t need to put the customer on hold and go searching for it. Those are some of the areas where I see massive positive changes happening within energy and utilities. From a challenges perspective, it all boils down to the non-deterministic nature of these AI systems. We need to have guardrails in place and ensure that evaluation is given the highest priority before even launching at scale.

18:04 – Mark
As you were talking — especially when you mentioned millions of customers — my mind went to Nissan as a global company with millions of customers across the globe. When you’re looking at AI solutions, can they be rolled out globally, or do they need to be tailored region by region?

18:24 – Zeke
There is always some level of customisation per region, per market — and by market I mean country, as we call it internally. For example, with voice of the customer and customer feedback analysis, globally we set up a framework that defines the broad categories we will be analysing. For example, sales versus aftersales, positive versus negative — those very broad concepts. Then regionally and by country, they decide how they want to manage it. At a broad level, they work within the sales and aftersales categories, and then some regions say they want to focus on dealers, while others say they want to focus on online or offline aftersales. So there will be variations. To manage that, we need to handle things a little differently. We’re still not at the final stage of this, but the vision is to set up a framework that allows them to do things for themselves, rather than global headquarters controlling all resources and serving them to the last mile, which is usually a little inefficient.

19:38 – Mark
Some of those use cases that Aswin mentioned require knowing what people on the ground need from AI. As we said earlier, AI isn’t a solution in itself — it’s an enabler for solving a problem. To do this well, we need to understand what problems we’re solving, and that’s easier if you’re closer to the front end. So I guess it’s about balancing how you capture the right things for AI to do, particularly in large businesses such as the ones you both work in.

20:07 – Aswin
Absolutely. I kept stressing that it’s not about AI or the fanciest models — it’s about what problems you’re trying to fix. It’s always problem first, no matter how much advancement we make in AI. You’d be surprised to know there are so many problems that can be resolved without any AI. You can have a simple rule-based engine, which is much more cost-effective, much more predictable, and you can still solve the problem. Having a fear of missing out and pushing AI into solving every problem is not the right approach. Something which really surprises me often is that the biggest gains come not from the fanciest models, but from fixing the fundamentals — like knowledge quality, consistent tagging, clear journeys, and clear handoffs. Those are very, very important. One of the ways we do that is by staying close to the front line — the customer service agents — listening to what customers are saying about us on social media, and drawing on NPS, which is one of the metrics we track very closely. Understanding the verbatim feedback — what are customers worried about, what are they complaining about — and then working out which problems are worth solving. Some problems are worth solving right away; others are not a significant issue for the majority of customers and can be addressed later. That balance is very important. Spending more time in the problem space before jumping to the solution space is crucial.

21:43 – Mark
Especially as getting AI right requires the right data as the foundation for all of these models and all of this learning. And sometimes dealing with that data is challenging in itself. So what needs to be in place from a data point of view before AI can actually start to improve the customer experience — before it complicates things and we start spending a lot of money when a simpler solution would have been better?

22:08 – Zeke
I can go first on this one. It’s actually a very common question — everyone wants to be data-ready before starting. First of all, data is of course the most important factor to feed into AI. In an ideal world, there would be no siloed data and everything would be connected, but that brings the first challenge — people ask about it precisely because it isn’t there. Then there’s the question of what the data means, how it was collected, how trustworthy it is, the volume and quality of it. Those are all questions that need to be addressed. On the other hand, the tricky part is that you can invest millions or billions into making data perfect, but it doesn’t generate outcomes or results unless you start to utilise it. The hard truth is: start now with whatever data you have and begin working on some use cases. As Aswin said, start with business cases that are worth solving and see what cost savings or value you can generate. Nowadays, the technology is good enough to solve most cases, and you can already reach a point where it generates some business value — rather than saying you want to build a perfect solution first, which requires data integration, data pipelines, governance, and security approval. Using 50% of the data to solve 30% of the problems is already progress — you’re getting 30% of the investment back. With current traditional machine learning, or generative AI for smaller volumes, the cost isn’t that huge. It will already help you. What I see in the landscape is that we’ll be moving to a stage where we have multiple different AI tools until we’re mature enough to have one AI platform that solves everything — which is unlikely in the near term — or until the data infrastructure has been cumulatively developed very well and we can connect everything together. But before that, it will be like the booming of the internet or the boom of digitalisation — multiple tools, and everyone will have a hard time governing them. But those will be the happy days.

24:26 – Mark
You’re nodding, Aswin. Do you agree?

24:28 – Aswin
I absolutely echo that. It’s going to be crucial, and I think we all face those challenges. Data is siloed in legacy systems in different formats. Beyond all that, what’s going to change a lot over the coming years is — I think the fundamental quality of the data is very important. And beyond everything, it comes back to: what problem are we solving? Based on that, get started with a proof of concept with whatever you have. You’ll never have perfect data. That’s only true in an ideal world, and we don’t live in an ideal world. We need to get started with a small POC, pick a particular use case, try it, and then iterate. And we shouldn’t treat these AI programmes as projects — they should be treated as products, and they should be iterated. We need to keep iterating and ensuring we have proper guardrails and evaluation frameworks in place to ensure that what we’re building really makes sense and is adding value.

25:25 – Chantelle
Just on the topic of data — drawing it back to the customer experience, I think as consumers we’re becoming increasingly aware of how our data is being used, and we value transparency from brands on how it’s being used. How do you build trust and transparency around AI in customer interactions?

25:43 – Aswin
When it comes to AI, customers are still very sceptical about it. Especially in energy and utilities, when you’re looking at multiple customer segments, there are some customers who are Gen Z and fine with it, and there are some who are elderly customers who don’t even trust digital technology, let alone AI. So it’s very, very crucial, as a brand operating in energy and utilities, to ensure trust is built with the customer. Because once they lose trust in you, they will not trust you back easily. It’s important that you let them know what kind of data you are using, and also ensure that they are aware they’re talking to an AI bot and not a human. Having that transparency across every segment of your operation and customer journey is very crucial. That is something which I think all the major companies in the energy and utilities industry are ensuring is taken care of, because ultimately it’s about customer experience. And if there is no trust and safety in what you’re doing, you will not get the desired outcome.

26:47 – Zeke
I totally agree with Aswin’s point. I worked on a global chatbot solution before for the airline industry. One of the key things we noticed is that you have to let the customer know it’s an AI bot. At that time, it wasn’t yet a hot topic, and people were confused — thinking, “Why does this human keep looping back rather than just solving my question?” Given the limitations of AI bots at the time, they started to feel they couldn’t trust the company anymore. The lesson learned: make sure it is clear to the customer who is serving them and what to expect, and always give them a clear alternative if they want a human agent. Even now, I still see a lot of companies trying to keep the customer within their AI bot loop, and then consumers just feel frustrated and leave, and the company board asks why utilisation isn’t very high. That’s a straightforward answer. Another thing is, when you’re collecting data, make sure you are aware of why you are collecting it and share that with the customer. Recently, a survey was sent to me — I think it was a national-scale survey about life situations, income level, and some very personal information. I trusted it had good intentions, but that information was not made available at the start of the survey. It felt a little too aggressive, and I didn’t want to share personal information with something I didn’t know how it would be used or stored, let alone whether there were cybersecurity risks. When collecting data, you need to ensure the customer is 100% assured that it is safe, securely stored, and that there’s no risk of leakage. Otherwise, the consumer will simply say no.

28:49 – Mark
I was wondering, Aswin — because you mentioned the benefits of AI in making the experience with a human agent in the call centre better, and you also mentioned different customers’ trust levels — are you in a situation now where you could almost know that a particular customer contacting you would be fine with an AI chatbot answering their query because they’re younger or whatever it is about them, whereas for older customers you’d say, let’s not even try AI — let’s get them straight through to a human and match them to the best agent for them given what their query might be?

29:25 – Aswin
That’s a good question, Mark. A lot of the time, customers do not understand how much their experience can be elevated using AI. At the end of the day, customers want their issue resolved or they want to get help seamlessly. It’s not about what technology is used — they don’t really care about that. We do need to be explicit about what we’re using, but we don’t segment customers, and I don’t think any of the major companies in energy and utilities do that either. What is important is the handoff. If the customer says, “No, I want to talk to an agent,” you should not keep persisting — “No, I’ll try to solve it, I’ll try to solve it.” There should be a seamless handoff where it is required, or if there are issues the bot cannot help with, we need to identify the intent and seamlessly transfer to an agent. That balance needs to be maintained. I never said it’s going to be all AI with no humans involved. It’s a combination of both, which is the winning formula for customer experience. That’s how I look at it.

30:23 – Chantelle
As a consumer, just to reiterate what you’ve said — if I engage with a brand because I have a problem, I will have a positive experience if they fix the problem. I won’t have a positive experience because they fixed the problem and they used AI. I don’t care how I’ve got to that point. So, looking ahead to the future, where do you think this is all heading? When we look back in a couple of years’ time at this transition, what do you think we’ll say about this moment?

30:48 – Zeke
I’m thinking back to two years ago when I first started looking at AI. I already assumed it would bring AI-powered superhuman agents, and I didn’t anticipate agentic systems emerging so quickly. When articles started being published about attention mechanisms and transformer architecture, I started to see there was huge potential in understanding more and more about the customer using the unstructured data we have, but I didn’t expect agentic AI to arrive so quickly. What I see for the future is still essentially an AI-powered superhuman agent. There’s a kind of argument now that one person is a team — one person with multiple agents is a team. That means we’re becoming more and more efficient from a company perspective. It also means we should be able to know more and more about the customer at scale, with a personalised level of interaction. One of the challenges is finding out how we reach that state. At this moment, there are just too many tools out there, each offering slightly different perspectives, slightly different modelling, and addressing slightly different problems. We’re not yet at the stage that CRM reached with Salesforce as a dominant platform, followed by HubSpot and a few others, or as ERP reached with SAP twenty years ago. We’re not at that stage, but I’m hoping we can get there sooner rather than later, so that we can leverage the latest trends and help the world be better.

32:21 – Chantelle
Aswin, what excites you most about this next phase?

32:25 – Aswin
Looking a few years down the line, what I would probably say is that this huge shift is actually from digital transformation — which was the big thing over the past decade — to decision transformation. Organisations that will win won’t be the ones with the fanciest models. They’ll be the ones who are able to redesign their customer journey around trust, governance, and measurable outcomes. The real innovation wasn’t AI as a whole — it was in how we transformed how we work and how we approach customer experience. What excites me most is enabling proactive, preventive service — resolving issues before customers even face them — and empowering frontline teams with context, clarity, and consistent decision support. That’s going to be a game changer. Of course, I also have some concerns. Rushing into scale without guardrails or evaluation frameworks is going to be a disaster for whoever does that. And using personalisation in a way that erodes trust is another concern. If a customer feels watched and manipulated, the brand damage outweighs any efficiency gains. Those are the two things that concern me. But overall, I’m very positive and bullish on how AI is enabling and transforming the customer experience in general.

33:53 – Chantelle
To close off our podcast, we’ve got a few more questions that may be a bit more personal than talking about your direct experience. So, obviously this is about customer experiences — as customers yourselves, what is the best example of an AI-powered experience that you’ve encountered? One that genuinely felt helpful rather than intrusive.

34:11 – Zeke
There is one good example. It’s a Finnish bank. I’m a loyal customer there — even though I’ve been in the UK for many years, I still keep that account. They offer a very helpful agent experience. It’s an inactive account that I’ve kept, and sometimes there are bills coming to that account that I’m not aware of. It will pop up a notification saying, “You don’t have enough funds in this account, and we’ve noticed you have another account with us that could cover it. Would you like us to proceed?” I click yes, confirm, and it’s done. That is a very smooth customer experience. Normally, in other places, you’d receive a notification saying you don’t have enough funds and you’re now overdrawn with interest. Then you’d have to log in to the bank, check what the bill is about, calculate whether it’s correct, and then transfer money and pay. But they manage it in a few clicks and off you go. I don’t need to spend more time on an inactive account, but they actually keep you satisfied. So I still keep the account, and I imagine they’re happy about that too.

35:26 – Chantelle
How about you, Aswin?

35:29 – Aswin
One example I really liked was Monzo. If you’re on a call and try to open the app, it automatically detects that you’re on a call and says they are not calling you — it’s a scam warning feature. Not only that, it actually helps you manage money in a much better way so that you can chat with it and make adjustments. That was a proof of concept they were running, which I was part of as a segment. It doesn’t exist in that form now, but it was pretty cool to see how much thought they put into customer experience and to what level they go. I was also part of a beta run for a leading travel website where, if you want to go on a holiday for a week to a particular place, it would work with you to understand your preferences, give you an itinerary, and go a step further to make the complete booking on your behalf. It’s a pretty exciting phase because, earlier, we used to search ourselves and choose. Now an AI agent will understand your needs, do the decision-making, and there’s no manual search involved. It can take care of everything. You still have checkpoints, but it was pretty impressive the way they handled it. What Zeke was saying about search being redefined by AI — where there’s no manual interaction and the AI can actually take care of everything — I think those two things really resonated with me.

36:49 – Mark
I had a good one with Monzo the other day — just a simple example. I ordered something before Christmas, it hadn’t arrived, and the company I ordered from wasn’t helping me. I had paid through Monzo. I’ve had banks help with this sort of thing before, and I went into Monzo’s help section. It said, “Ask me a question.” I said, “I haven’t received this order from this company. Can you help me with this?” The AI-generated answer was so precise — “Yes, we can. Go to your account. Find the transaction. Click on ‘Support with this transaction.’ Go down to ‘Dispute this.’ Upload these documents.” It was so accurate and concise, and I knew exactly what to do. It was almost as good as — if not better than — having someone talk me through it. It was really great.

37:33 – Chantelle 

Very interesting.

37:34 – Aswin
In the old days, they would just push you with a link — click on the link and find it yourself.

37:39 – Chantelle
Yes, or having to be on the phone to someone for a long time. I think historically people valued hearing someone’s voice on the other end, but because people lead much busier lives now, we value convenience a lot more.

37:53 – Zeke
Totally agree. At one of the events, I was in a discussion with an expert who claimed that people prefer to stay on a call because they want the personal connection. To me, that’s a challenging view. It seems like they’re not realising that people are changing — consumer behaviour is changing.

38:11 – Mark
So our final question — we ask everyone this. I’ll start with you, Zeke. Who has been the greatest professional influence on your career, and what did you learn from them?

38:20 – Zeke
That’s a good question. One of the good influences I had early in my career was Marco Lauka. He’s the Managing Director of the Salesforce unit at Accenture in Finland. I spent two very happy years there. When I left to work for Finnair as a manager, he told me, “I have no doubts about you becoming a future leader. The only thing I’d ask you to pay attention to is to develop the people around you and below you — in an organisational sense, the people who report to you. Develop them so that when the time comes and there is a new opportunity, you know you can hand over whatever is on your desk to someone you trust, and they can get things delivered. That’s how you move forward.” I keep thinking about that all the time. It’s actually why I now have a more open mandate for collaboration rather than saying, “This is my area and I don’t want anyone to touch it.” I’m always saying, “I can help you, and you can help me — let’s do that.” And I think that’s the right way to do business.

39:27 – Mark
Absolutely. That’s great advice.

39:29 – Chantelle
Great leaders empower others.

39:31 – Mark
For sure. Aswin, your greatest professional influence?

39:35 – Aswin
The biggest influence for me was from a leader and mentor early in my career who pushed me to spend time with frontline teams and customers before even writing a product roadmap or thinking about building something. He always used to say, “At the end of the day, you’re trying to solve a problem, and you need to spend enough time in the problem space to know you’re doing the right thing.” I learned that the best strategy often sits in the details of real conversations — understanding what market you are after, what customers are facing, framing the problem properly and clearly. I think that’s half the work done. Then you move into the solution space and figure out how to enable that. That stuck so firmly with me that since then I’ve always looked at the problem first, which has helped me to build impactful products that create genuine outcomes. That’s something which really stuck with me and had the biggest influence.

40:36 – Mark
Well, thank you very much, both of you. I think that’s about all the time we have. It’s been a great discussion. It’s always fascinating to talk about AI and hear different points of view. We really appreciate you coming and joining us, and thank you for all of your input.

40:39 – Aswin
Thanks for having us.

40:40 – Zeke
Thank you very much for inviting me.

40:43 – Chantelle
Thanks, guys. That was a really interesting conversation with Zeke and Aswin. I think the underlying theme was that AI is an enabler — it’s not always the solution to a problem. You have to approach customer experiences with the problem you’re trying to solve for the customer in mind first.

41:17 – Mark
That really came out when we asked them, “In two years’ time, when we look back on this, what are we going to be seeing?” Neither of them talked about AI replacing what we do or AI doing roles that people are doing today. They both talked about AI augmenting people. Zeke said it’s about superhuman interactions — it will be a person with a team of agents. And then Aswin talked about reorganising businesses and transformations, but with AI helping you make decisions so you can organise your business better and serve customer needs better. So very much it was nothing about replacing — it was all about improving.

42:00 – Chantelle
Complementing.

42:01 – Mark
Absolutely. It’s great to hear different points of view. It’s great to hear the way AI is being applied from two people who clearly know a lot about it and for whom it’s not just a buzzword. You can see they’re really building solutions for their businesses that help them go further.

42:16- Chantelle
Thanks for listening to our podcast. Please share if you’ve enjoyed what you listened to today, and follow us for next month’s episode.

42:23 – Mark
We’ll see you next month.

42:24 – Outro
The CX Equation is brought to you by TapCXM. To find out more about what we do and how we can help you, visit tapcxm.com. Make sure to search for the CX Equation on Apple Podcasts, Spotify, or wherever you usually find your podcasts. Make sure to click subscribe so you don’t miss any future episodes. On behalf of the team here at TapCXM, thank you for listening.