From Assistants to Agents: The Future of AI in Financial Services

Speaker A: Foreign.

Speaker B: Hello and welcome to the latest episodes of the deciphered podcast by Dane and Company. On this podcast we unpack um, the stats to give you an in depth perspective on um, different topics relating to fintech and the financial services industry. I'm your host, Jeff Thiesse, global head of Fintech at Bain Co. And for this episode we are recording live from the US edition of Money20 20 in a relatively sunny and hot Las Vegas, the annual fintech mecca. It's always a pleasure to be back in my beloved Las Vegas. I probably have spent too much time in Vegas in the last 15 years or so, but hey, that's one for another day. Uh, in this episode we'll explore one of the hottest topics in fintech right now which is AI and agentic commerce and the impact this will have on financial services. I am super excited to be joined by my colleague and co host, the one and only Mike Cashman, co head of our global fintech investor practice and similar to me, a Vegas regular. By now I believe this is what the third or fourth was the time that we're co hosting a live show in Vegas together. So this has become a bit of an annual tradition. I am also super excited to be joined by two phenomenal guests. So let's introduce the people around the mics, starting with Mike Edmonds from PayPal. Mike, how's Vegas treating you so far?

Speaker A: It's good, it's good. You know, it's a lot different this time of year than being in Chicago, so I'll say I'm enjoying the sunshine. The few times that I get to actually step outside and prove that there is the sun actually shining, it's been great, honestly. In ah, all sincerity, this is my first money 2020. Ah, well, which is pretty wild. Um, I'm really enjoying it. It reminds me of fintech version of NRF in New York, which I have frequented. Um, but really happy to be here.

Speaker B: Awesome. Yeah, it is nice. I do need to carve out more time to actually go outside and see some daylight and get some fresh air. Supposed to be in a conference at the Inter. Very good for your health, Mike. Please also tell us a little bit more about your role at PayPal.

Speaker A: Of course. So I joined PayPal just about three and a half months ago. I lead agentic commerce from a commercial growth perspective. What that means is I'm informing and leading our product strategy, the different product bets that we're making. Um, I'm accountable for revenue growth, for customer acquisition and also organizational and operational alignment. Before joining PayPal. I led AI strategy for Microsoft's worldwide retail and consumer goods industry team. And a little fun fact about me back home in Chicago, I'm an adjunct professor at Northwestern where I teach a, uh, course on human center design and product management and also business models.

Speaker B: Awesome. Well, we are delighted to have you as a guest, Mike. I am also very pleased to be joined by the fantastic Pahal at Nvidia, where welcome. What's been your personal highlights of Money20 20 so far?

Speaker C: Thanks, Jeff. It's been a pleasure to be here. And the highlight has been the ecosystem being at uh, money 2020 all these years. And what resonates is that the things you talk back at work in today's world, it's agent ecommerce, it's foundation models, it's embedded in finance. Uh, the same themes are resonating here as well. And that is super exciting to be at the show and get that validation from the ecosystem.

Speaker B: Very good. And obviously the listeners will be familiar with Nvidia, but tell us a bit more about the stuff that you focused on it Nvidia. I know you do some really, really cool and exciting stuff as well.

Speaker C: Absolutely, yeah. Nvidia as a company, we are known as a chip company to begin with and a lot of people perceive us of that. But what's growing as a part of that story for the last decade and a half is the full Stack computing platform story, the AI platform which enables the applications which you hear day in and day out from the broader ecosystem. And as a part of my remit is to marry that horizontal platform, that Full Stack platform with, with the broader vertical ecosystem and the use cases that come as a part of it when it comes to AI. So I manage the global practice for the payments industry and as the payment companies and the firms build upon AI use cases, how do we bring together Nvidia's platform to ultimately propel those into the right direction, get the KPIs to life with AI being at the center of it.

Speaker B: I couldn't think of two better guests to join us for this conversation. Then there, guys. All right, let's move on to the show. Um, at Bain, we sometimes use a concept called an answer first, which is essentially having a relatively formed quick hypothesis using the collective facts, stats and brains from Bain. We're going to adapt this a little as we always do on the show and ask our panelists for your quick 60 second answer. First to the top line question, which is how is agentic AI set to transform the financial services industry? Mike, I'm going to come to you first on this one.

Speaker D: There's two things that I'm most looking forward to. One is going to be much more personalization and scale and a reduction a lot of frictions as consumers interact with financial services. I think the second and a little bit different is going to be real proven out use cases for AI that work in financial services versus a lot of the experimentation we've seen over the last three years.

Speaker B: Very good. Um, paho.

Speaker C: Absolutely. I mean in the very simple term it will take the boarding out of financial services. And what that means is that this industry has been laden with workflows, it's been laden with processes. And with agentech AI coming to helm, there's a greater opportunity to supercharge these workflows and convert them into something more actionable, something more dynamic. And that is what would define the next frontier for Agent Ki and financial services.

Speaker A: Love it. I'm going to take it from an agentic commerce perspective across industries at large and I'll say what's going to happen is agenta commerce is going to completely reimagine the way that we shop as consumers, but also the way that we connect with and reach shoppers as merchants. So think of it as both sides of the market. I'm um, incredibly excited. Mike, you mentioned the word personalization. So I think through agents e commerce, not only personalization but also truly individualized conversations like down to the one to one I'm really fascinated about. And then from a merchant perspective, just think of a net new channel, the ability to reach new audiences in new and different ways with your products and services in a safe, secure and seamless fashion. So very excited about it.

Speaker B: Very, very good. Definitely, uh, lots of hype and excitement also just walking around the show floor. It's funny how two years ago we didn't see that many crypto companies and now all the crypto companies are back. Everyone is talking about AI and agentic commerce. We've seen lots of exciting announcements. But actually uh, I want to build on the last point a bit Mike is when we talk about agentic AI, what do we actually mean and how does it differ from the AI that most financial institutions are already using?

Speaker A: So I'll make the distinction between traditional AI gen or generative AI and agentic AI just broad stripes. I would love to hear Paul, if you have a uh, build on it. But the difference between traditional AI and generative AI, think of the difference between curative and creative AI. So curative think of traditional AI in a sense of deterministic algorithms identifying signals and insights and driving individual outcomes. So think of if you're on your favorite social media platform, something like LinkedIn. LinkedIn, you know, based on who I am, who I've engaged with, the things that I've liked, the things that I've commented on, curating down into content that I'm most likely to engage with next. Like that's one example of traditional generative AI is creation of AI. So think of humans interacting with the most sophisticated technologies of our lifetime, which are large language models, whether it's through text, image, video, voice, input, output, the ability to say or ask a question and have those models create net new content or new conversations that have never existed before. As we get to agentic AI, think of what really the anatomy of an agent, knowledge, tools and memory. So this is about equipping agents to act either with humans or on behalf of humans that are grounded on specific data. So they're intelligent around the data that it's grounded on. They're given and equipped with tools to take action. So it's not just like providing an insight, but could be actually completing a workflow and then from a memory perspective, this is both short term and long term memory. So the ability to learn and to continuously get better over time.

Speaker C: Absolutely. No, I mean I would add a timeline to what Mike alluded. The big thing is that the early 2000 and tens was the era of prediction and you were predicting anomalies and images, you were predicting anomalies and your transactions, et cetera. And you're still doing that today. But the early 2000s was the era of generation and retrieval. You're retrieving from knowledge bases, you're retrieving information and fetching it to the nearest accurate relevant context, if you will. And today's era is agentic era where you are adding on top of it the action layer, the reasoning layer, the thinking layer. So just how humans would make a decision, we would think deeply, explore our options, ultimately evaluate, plan and act. That's the agent layer which is overlaying on top of the generation and the retrieval pieces. And what's phenomenal in this piece is that in the agentic era, uh, the number of tokens, the number of sources which these models have to go and fetch is humongous. You're not just talking about one or two sources, but rather hundred sources to form a viewpoint, do the research and ultimately build that action plan and take a decision on behalf of the user.

Speaker B: Yes, I couldn't agree. I had a really conversation with one of our clients the other day about the shift that we're seeing in the industry from banks about effectively being pretty reactive to trying to become more predictive and more proactive to your point, because obviously there's so much data that CL banks have. But the question is, how do you utilize that in the best possible way to put consumers more in control? So I'm similar to you, really, really excited about how this space is going to evolve. And it's still relatively early days. Mr. M. Cashman, why do you think this moment represents such a turning point for financial services?

Speaker D: There's one thing I analogize this to for financial services, and that would be digital and mobile banking, where we started to get rapid access to financial services to amass amount of consumers. It was much more real time. And I think what has happened in digital banking is it's had a barbelling effect. Right. And so the largest banks have benefited from this as well as challenger banks in the middle. It's been a lot tougher, Right, for a lot of local community banks, for a lot of the regional banks as well. I think that this is going to exacerbate that in the future. And so being heavily digital and willing to experiment and experiment at scale, including in a highly regulated environment is going to be what it takes to actually continue to succeed, both in making sure these actions are useful, but also that they're seamless. Right. For the underlying consumer. Because I actually don't think that the consumers need to know how it works at all. And I suspect we might not even be using the word agentic in a year or two. Right. I think that word is descriptive now because it's almost experimental in the way that it's happening in the future. It's just going to be done the

Speaker B: same way that no one really talks about digital normal banking before, even though there's still a fair few banks around the world, uh, that have a long way to go when it comes to creating a much better customer experience.

Speaker D: Exactly. Exactly.

Speaker B: Right.

Speaker A: For sure. What I love about that too is how we're excited about it as experts in the field. But to your point, the people on the other end just want it to be done. And that's, I think a fascinating part of the equation is is it the difference between traditional or generative or agentic? I think at the end of the day, that's really not the point. It's not an or, it's an and it's how do these things come together to create outcomes, to drive experiences that are truly transformational Yeah, I remember whether

Speaker B: this was at Money 2020 or another conference, it must have been eight, nine, 10 years ago where people talked about this concept of self driving money or invisible money. So now is this a different flavor or are we just recycling uh, and coming up with new terms? But from your perspective, and again, I know you're speaking to a lot of different financial institutions as well, why do you think this is such a turning point for financial services? It's such an exciting opportunity.

Speaker C: Absolutely. If you think of the genesis of the financial services institutions, there are two key pieces here. The data stored in these institutions is all tabular, which has been used for insights all this while. On the other hand, there's unstructured data in the form of PDFs and um, sharepoints and all that that was just sitting there. And there was lack of technology to harness it and bring to life in an insightful manner. With generative AI and agent AI, those possibilities are opening up. And the second piece is when you take all of that information and overlay onto your workflow, which were previously you would call process automation, etc. And the core banking workflows, the core institutional flows have been working step one, step two, step three, then do this, if not, then do that. Those are very static workflows. They with time can become inefficient and at the same time just puncture the entire organization if you will, at a micro uh, level. And that's the opportunity with Agent Aki brings that. How can you build fine tuned niche specific agents that this agent would do particular task around reconciliation versus this agent is very good to understand the W9s versus that payslip you uploaded onto your mortgage application could be verified by so and so agent. So these fine tuned agents would be the future of what would aid the workers and the employees at these organizations. And that is m something very powerful which a lot of institutions are yet to fathom.

Speaker B: And just in the last couple of days Here at Money20 20, we've seen a lot of exciting press releases and announcements via partnerships. But where would you say we are on the adoption curve, Mike? Are, uh, most institutions still experimenting or you know, starting to see real implementation and adoption at scale?

Speaker A: Yeah, uh, the way I talk about it, we're in the early to middle innings of the adoption curve when it comes to agentic commerce. I mean, agentic commerce is actually truly in the early innings. I think agentic AI probably getting closer to the middle. But in terms of we've all been around these major computing shifts. Right. And always in the moment it feels like this is the biggest, the fastest, the craziest. Yes, I can actually say, I mean with all honesty, this one feels fundamentally different.

Speaker B: I agree.

Speaker A: So even just a month, two months ago, where it's literally just about testing and learning and experimenting and finding ways to measure and how will my shoppers on the other end actually interact in these new channels? We're seeing such a rapid adoption where people are moving forward at scale. It's not if, it's when. And this is an opportunity where whether it's retailers or consumer goods companies exposing their inventory across large language model surfaces to reach this new audience and in that new channel or adopting agent e commerce technologies on their own surfaces, we're seeing this move at breakneck speed.

Speaker D: Ah.

Speaker B: And we'll talk a bit more about some of the challenges when it comes to really driving adoption at pace and at scale across the organization in a second. But before we do that, we touched some of this earlier. Mike, but what are some of the most promising use cases that you're seeing across banking, payments, investment management and other parts of the FS value chain?

Speaker D: You know it's interesting because I think the most notable instances we're actually seeing and observing are almost all around commerce to date. I mean just from the end count of activities that you do with financial services, by far the single largest activity is buying. We have seen a whole bunch of initial experiments across this. What we haven't really seen yet or at least publicized yet is anything that's going to go really deep into the core offering essentially touch the core money storage or investment management. I think we're likely to see that evolution happen strictly starting in commerce and starting in the way people are uh, taking that action of I have this need and then gets fulfilled the action without needing to take all of the steps in the interviewing.

Speaker C: I'll give an industry spin on that. Particularly what we are seeing with customers like Capital One. They have a um, chat concierge on their platform and one of the features in that is an auto navigator. Now any user who would want to buy a vehicle for the first time has to navigate a lot of complexities. Especially in the US is where I'm very familiar with is the research is a lot of tabs and understanding a lot about financing options etc. Gets overwhelming. And uh, what they have done is they have built specialized agents which one understand the risk profile on your side on one half on the other side it understands what are your options basis at risk profile, researches that researches the Internet for you, plans Accordingly and gives you a set of actions that, okay, this could be your, uh, path one, part two, part three. So that gives a lot of satisfaction to the customer at the end of the day, kind of aiding them in the process and making them feel more homely. Ultimately, that drives loyalty and stickiness from a customer perspective. The other example which we have seen is the likes of rbc Royal bank of Canada. They have a platform called, uh, Aid and Assist. And what they have done is that they have armed their private bankers with agent AI capabilities. So if I'm a private banker, I, um, am, uh, consulting my customers at the end of the day, if there are any macro changes which would affect their portfolio. So how could agents handle that trigger and understand what is the impact on the portfolio performance? So instead of these bankers going and sifting news from all over, it's agents aiding these bankers who are ultimately consulting the customer that, okay, these could be the few parts of actions which you should take.

Speaker B: One of the things that we talked about earlier was the shift towards autonomous agents. Again, this whole concept of self driving money and putting customers more in control. Personalization, Mike, based on the conversations that you've been having, you know, all of this effectively redefined customer engagements and really moving beyond personalization to genuine autonomy in serving those customer needs.

Speaker D: Yeah.

Speaker B: And one of the challenges that we still see in financial services is this one size fits nobody approach.

Speaker A: Yeah. So I'll answer that question through the lens of some of our big news that we announced this morning because I think that really perfectly timing. Yeah, there you go. There you go. Yeah, thank you for asking that question. So we're incredibly excited to have announced our partnership with OpenAI this morning, which is transformational as we are working with OpenAI to expand the AgentA Commerce Protocol ACP to first and foremost include wallets, starting with the PayPal wallet. Really excited about that. And in the future, we're going to be continuing to collaborate to expand and have the Venmo wallet show up. Why that matters is that consumers who have the PayPal wallet will have access to any of the funding instruments that sit in that wallet. So think of like banks, balances, rewards, crypto, buy now, pay later cards. You know, the list goes on and on. Number two, it's about our collaboration with OpenAI is empowering our merchants to be able to surface their products, their catalog, their order management system on instant checkout, to really have control and agency, for lack of a better word, over this new channel, this new growth channel, and then that's through a product that we just recently announced today through our gente commerce services portfolio called Storsync. It's the ability to integrate what you have showed across many different agentic AI surfaces. And then lastly, we're going to be providing payment services for card processing and that's inclusive of Apple Pay, Google Pay on Braintree, the PayPal commerce platform, as well as other gateway integrations. So we're really excited about that because these kind of partnerships make agency commerce real, right? This is bringing these new capabilities, these new experiences, these new, you know, end to end flows into the hands of shoppers around the world. We're really excited to be a part of it.

Speaker B: Very cool. You guys have been very busy here.

Speaker C: I know that's been fantastic. And to top off the partnership is something coming out of the press just an hour ago.

Speaker A: Go ahead, Paul.

Speaker B: Uh, come on, come on, come on.

Speaker C: Vroom, vroom, vroom, vroom, vroom. So PayPal and Nvidia have partnered to bring the power of open source foundation models that power these three zinc capabilities and the conversational search capabilities within Asian tecommerce agents to ultimately enable the broader SMB ecosystem. So PayPal has tens of millions of merchants on the other side of the platform. How could you enable those SMBs to ultimately be armed with uh, these technological capabilities that coming out of the innovation which we are seeing? So they have been great proponents of a platform and working together, we want to bring this power of open source technologies and just be abstracted and embedded because at the end of the day, the SMB merchant wouldn't know what's the tech powering behind it, but the benefits that come with it, in terms of more personalized offers, in terms of just seamless checkouts, et cetera, is something what gets them going. And that is a very great partnership that we have with the likes of PayPal.

Speaker A: Very excited about it.

Speaker B: Congratulations to both of you. We timed this podcast in such a way so you had an opportunity to get the announcements out of the way. So this is perfect, perfect timing, guys. Hey, I wanted to come back to something we talked about a couple of minutes ago, which is some of the barriers around adoption. And Mike and I have had lots of discussions with our clients and we've spoken to many boards and executive teams around the world. And one of the key challenges that many organizations face, not just in financial services, is how do we go from doing endless proof of concepts to actually deploying this at scale across the organization. How do we drive adoption within the organization? Because one of the challenges you can give a 30,000 person business, you can give everyone the latest tools in front, but driving adoption is a very, very different story. I had a really fascinating discussion with the group head of AI at one of the largest banks in the world. And it was a bit like the wild west where every part of the business was sort of doing their own thing. There was a complete lack of consistency in how to go about building some of this stuff. So it is quite difficult also, given how fast the space is evolving for organizations to really benefit from all the amazing things that AI has to offer. Mike, from your perspective, what do you consider to be some of the biggest barriers to adopting? Is it technological, organizational, cultural? All of the above.

Speaker D: I think right now we're seeing a lot of experimentation still across the board. I think what we haven't yet seen and I think there's going to be plenty of opportunity to do plenty of fast followership here in this market. And so once something takes off, the barrier to actually copying that is actually going to be relatively low, which will reward scale for some of the providers that can actually move pretty fast. I think, uh, a lot of the barrier for many firms will be both organizational as well as risk. Posture wise. It may not actually wind up being true risk, but the perception of risk and the desire to slow things down, put them through the checks and balances that a bank or financial services institution typically has, these are pretty significant blockers even when they don't have to be. Sometimes they're truly regulatory in nature, oftentimes they're not, but they're actually more cultural and organizational. And so I think that that is going to be the probably the single biggest rate limiter on a lot of firms. And I think that's one of the things that's going to separate the people who jump out ahead from those who don't.

Speaker C: I love how Mike uses the word rate limiter sits in this space.

Speaker B: Yes, he does know his stuff.

Speaker A: So plus one, uh, on all things cultural, at the end of the day as fascinated we are about the technology itself, it's about the people and the right roles with the right competence to deploy it. I will say what I'm seeing is interesting between my time at Microsoft and then into my time at PayPal is shifting into PayPal all focused on commerce. These pilots look different when you're actually releasing new experiences to live customers. And my time at Microsoft was the earliest innings of AI where this is around different pilots and different business units and different orgs and people kind of testing things out, picking them up, running them and then literally like putting them back down. When you start to do pilots in the world of E commerce, you're playing with live data, right? So now when we're unlocking new channels, whether it's ChatGPT's instant checkout or Perplexity or Google or Microsoft Copilot, the pilot is how are your customers actually using it? Which metrics is it actually moving? And are these channels that you continue to double down and invest on or pull back from? I'm excited about that because as a product person, like that's when you release these things into the wild, that's where you really get the true sense. Like is this really making our business better? Is it improving the lives of merchants, customers alike? So excited about where that's going.

Speaker B: Yeah, no, very true. I had an interesting chat with someone yesterday about how uh, they sort of ditched the whiteboard and gone straight to 5 coding just to accelerate the process actually getting something into customers hands as quickly as possible. But from your perspective, what do you see as some of the biggest barriers?

Speaker C: I think they're outlined really well. It boils down to data culture, you know, having your platform ready to be successful with AI and the classic uh, framework to you know, go get there from zero to success in AI it would be like start with your use case, you know, be very focused and get going. See early value and then scale, you know, classic framework that's kind of hampered to a certain extent thanks to the progress which we have seen in the AI space. And every new day there's a new model, there's a new protocol, there's something new coming in which you know, banks and financial institutions would like to take a step back and then think further. So that scares them. And um, the way to mitigate that and which we have seen with our large customer and institutions is that start internal, start small. Some of the use cases around agentic AI for software engineering, for your QA testing, be an aid to your employees in that sense and then continue to see where you are having those blemishes and improve upon. That'll give that institution enough confidence that hey, maybe tomorrow I can release it to the entire internal base and then day after tomorrow I could release it externally as well. Obviously comes with guardrails and the human in the loop aspects of it. But like that's the path and the way to think about it.

Speaker A: Pahal what I love about what you're saying too is like one word to take away from the whole barrier conversation is trust. Trust internally. Do employees use it trust it from an external, from a shopper perspective, can I trust the agent to work on my behalf? So I think that's when it comes to culture being really specific about measured outcomes and just always on this foundation of trust I think is really important

Speaker B: for just to build on that. Obviously trust is a critical issue in finance. I've heard lots of people talk about uh, the challenges when it comes to establishing trusts, uh, considering all the exciting stuff we just talked about. So how can institutions really build and maintain trust in AI driven decisions, whether that's credit underwriting or whatever it may be, especially when customers might not fully understand how those decisions are made. And I had uh, an interesting chat yesterday where someone said well if I basically have this Jeff bottom that is almost running my financial life and I had two bottles of wine and I allow my bot. So I there's a notification on my phone that says do you want to invest xyz And I'm like yeah, please go ahead. And I lose a lot of money in the next couple of days like who's actually liable? So obviously there are some issues uh, that we still need to work through. But Mike, give me your perspectives on the topic of trust.

Speaker D: Yeah, I think what's going to happen is we're going to get to some of those guardrails and go past them in a couple examples pretty quickly and there's going to be mess ups that happen along the way. I think part of that's going to, to act as even more of an impediment for incumbents to continue to push forward. I think this is going to play partly into the hands of those who are moving a lot faster, a lot smaller and frankly who have less to put at risk from a liability perspective, from a regulatory perspective. But you know, we're going to see transactions go awry and we're going to see investments go awry. I think what, what actually is going to happen though and we've seen this in other examples that put out like Neobanks as a good example is when there's enough value being provided to a wide enough array of consumers writ large, I think the rules have to start migrating towards where this is going. And so when you start to see truly better financial services outcomes, truly better offers being presented at the right time, at the right place and taking frictions out of the system, I think we have to start weighing the cost benefit of that in a different way. And what I liked about what Bahal said earlier is getting the niche personalization and so really unlocking if you think about what open banking's promise was, I think a lot of it actually hasn't really come to fruition. And so things like delivering the right loan to the right person at the right time at the lowest possible rate, optimizing the yield on deposits in the right way, getting the right product into their hands. There's a lot of inertia in financial services and so being able to change that even on the margin is enormously beneficial to the consumer.

Speaker C: I want to double down on that. Particularly what I take away is a segment of one conversation. Today you cluster customers into groups but the moment you are able to deliver that segment of one which is obviously far out from where we stand today. But there are efforts in this space, particularly how AI is aiming is super important. There's a new concept called payment foundation models, open banking data. You bring all your transactions so you have that entire Persona of a customer embedded into those transactions. The models today which underpin generative AI have the ability to encapsulate those static behaviors and those dynamic behaviors about a customer so they are able to predict better and build that customized plan offer, et cetera and that's next best action with respect to what this customer would gonna be doing. And I think that is a very powerful message. Also ties into the world of agent E commerce. Commerce because personalization would be a big aspect of it. That's where it would move the deep up there.

Speaker B: We've got a couple of minutes left guys. I wanted to future gaze a little bit again. Mikey, uh, mentioned earlier it's early days, right. We're in the early innings at the moment and even though there's a lot of excitement, there's still a lot of work that needs to be done to properly scale this across the industry. So how do you guys expect Agentica to really reshape some of those competitive dynamics in financial services over, I don't know, let's say the next year. Three to five years?

Speaker A: Yeah, I mean I think from a commerce perspective, I mean three to five years feels like an eternity, right?

Speaker B: Not quite. Yes.

Speaker A: Three to five months.

Speaker C: Yeah.

Speaker A: Things are looking a little bit different. I mean I think over the near term we're going to see these agents E commerce channels take shape as viable ways to reach and interact and deliver these segment of one type relationships with customers. And I think we're going to see that groundswell of larger and larger percentages of E commerce go through agentic channels. Now agentic channels are going to change by their nature. Right now we're talking about checking out on agentic surfaces which still humans have, you know, both hands on the wheel. But it's still a net new channel. I think we're going to get to a world very, very quickly where now humans are delegating permission and authorization to agents to do things on their behalf. I don't think over the near term every product category is going to be completely autonomous because at the end of the day, shopping is very much a human centered. They're always talking about a new outfit, you know, especially things like luxury, high consideration items. But I think as we continue to cross that trust barrier and get more and more comfortable, I think more and more product categories will be influenced by Gen 2 commerce. And I think a greater share is going to move from traditional channels to the new ones. Now in five years, I think we're talking about like a lot of vision, like completely multimodal what's happening in your brain and how that translates to what you want. I mean that's really how quickly these things are moving. Yes, but I think over the near term it's about this rapid adoption of the channels that we know and the agents doing more and more on our behalf.

Speaker B: I think they're absolutely spot on and it is fascinating to see how quickly this space is evolving. It's hard to keep track of everything that's happening in this space. Right. But, uh, one question for all of you. Again, uh, there's so many organizations who are desperately trying to figure out what does this mean for us as a business? What are the implications, what are the opportunities? Is this a defensive play? Is it an offensive play? If there's one piece of advice that you could give to your CEOs or people listening to the podcast or other businesses that are trying to figure out again, what does this mean for my business? What would you say to them?

Speaker C: Absolutely, be prepared and be ready for what's to come. Because things are always going to be dynamic. But it's the investments you would make with a longer vision. Certainly there are areas which companies decide that, hey, I need quick ROI on this. But like the space is moving so fast that it has to be a hybrid, measured approach where you would see the benefits of agentic AI coming in in the long term. So I think that mindset is important to have research as well as your revenue generation capabilities going hand in hand.

Speaker D: I think the one thing I'd say is make sure you're actually having the right sensing capabilities in your organization to see what's getting traction the fastest in your space. You know, we're still at this period of experimentation. We're still in early innings of actually deploying this to consumers, to businesses. And some of these use cases I think, you know, probably would all suspect commerce is likely to be one of the first and one of the fastest growing. Um, but having the ability to A, participate, B, set up your organization and then C, very, very quickly enter, uh, copycat, take best practices from others as quickly as you possibly can is going to help keep both the organization sharp and fresh. And it's really at this point sometimes only going to require a, uh, modest investment today to make sure you're actually on top of it for the future. It's been a couple of years now since we've been talking about agentic commerce and agentic AI at Bain, but it's only in a couple months since we've actually started to see this happen in the wild.

Speaker A: Yes, I would say now's the time to play offense. We talk about this a lot at PayPal. I think there's to do so in a way that's pragmatic, that's responsible, that's measured. But I'm really thinking about even this holiday season into Q1, as those who are on the bench versus those who are in the game are going to be in such a different position come the turn of the calendar year. I mean, your ability to leverage in a, even in a, you know, narrow, whether it's just a product category or one specific surface, your ability to be one of the few who are going to push forward, test and learn in these new channels as you get into next year. I mean, the gap between those in the game, those on the bench, I think is going to be significant.

Speaker B: I could agree more. There's one thing that I would like to add which is get your executive through the CEO of the business, set an example. So get your executive team to actually play around with some of this stuff. But there's a very large bank actually forced all of their executives to go and build some of this stuff, provided them with the tooling because don't just rely on your innovation team to drive some of these initiatives or your head of AI, uh, to drive adoption. I think it's such a fundamental game changer for the industry, for your business. And therefore I think if you're a senior executive, you have just as much as a responsibility to play around with some of this stuff as well and really set the example for the rest of the organization. Why?

Speaker C: I got it.

Speaker B: Exactly, exactly. I love all these new terms. I vibe coding, uh, Guys, that was a fascinating discussion. We could have easily continued this conversation for another 45 minutes, which we'll definitely do over some nice cold beers or a nice glass of wine, which I'm desperately looking forward to. Um, thank you so much for listening to this episode of Deciphered. Where can listeners find out more about you? Mike?

Speaker A: Yeah, LinkedIn is the best place. Mike Edmonds on LinkedIn, that's where you see the latest and greatest updates, all the different partnership news and announcements.

Speaker B: Very good.

Speaker C: Same for me. Quite active on LinkedIn.

Speaker B: Very good, Mr. Cashman.

Speaker D: You can find me on LinkedIn and on Bain.com, i was guessing at the

Speaker B: end, no telegram or. Yeah, any. No, not yet. Give it some time. The easiest way to find me is on LinkedIn as well. We could look at all of my musings and of course@, uh, bain.com where you can find a lot of really great and exciting content. Um, if you enjoyed that, please leave us a preferably 5 star review on Apple, Spotify or your favorite podcast platform. It's does make a difference. And also please subscribe to the podcast so you never miss an episode going forward. Mike Bahal, Mr. Cashman, thank you so much for joining us in sunny Las Vegas. Let's go and get a nice cold beer, get some daylight, get some fresh air for listeners. Thank you so much and we'll see you next time.