Competing for Recommendations: The AI Midyear Pulse Check

Speaker A: Foreign.

Speaker B: Welcome to a brand new episode of One Vision. This is Theo, your host for today's episode. Now it's halfway through 2026, so as we have been doing in the past, thought we would bring everyone back in for a quick recap of the year, especially since Vinovate Spring just wrapped up and see what's ahead of us. So joining us today is our wonderful friends, Julie, Tiffany and Jennifer. Welcome, ladies. Thank you.

Speaker C: Thanks. Hello.

Speaker B: Yay. So we, um, had a lovely time in San Diego, Julie. Like I don't even know how many times I've said it. Thank you so much for moving the conference to San Diego. We just love it. It's beautiful, it's gorgeous and everyone is happy. Um, let's start with that then. What caught your eye from the recent, uh, three days in San Diego from people that are demoing solutions, things, uh, that we're talking about, exhibiting keynotes, et cetera. Did anything catch you by surprise?

Speaker C: Yeah, I wouldn't say I was necessarily surprised by anything. Um, but one thing we did do differently in uh, Fundabate Spring in San Diego this year was we had, um, Impact plus sessions, uh, which were a lot of fun. It was kind of like taking the demos but shrinking them down to four minute pitches. And instead of, um, you know, viable technologies, they are very early stage companies. So, um, those were a lot of, um, kind of, yeah, the early stage concepts and of course most of them were centered around AI. But as far as the regular demos on the main stage, I think one thing that has been trending both with, with Finovate Spring and then Finovate Europe earlier this year. I've noticed that AI is no longer being just tacked on to, um, existing platforms, uh, like a customer service, you know, chatbot, something like that. It's now being integrated into solutions and that it's really kind of happening without. And of course AI is everywhere, um, throughout the demo stage. I'm not sure if there was a demo without AI on it. Um, I didn't get to sit on all the demos. So if there was one without AI, I missed it.

Speaker D: It wasn't the kick.

Speaker C: The mouth is really being integrated into all the tools pretty much.

Speaker D: And I would say I saw some of the same themes that we see in our research. Right. It's really all about AI fraud, personalization and even customer experience. I think the biggest thing that stood out to me this year is how practical the conversations have become. So we were focused before on just trying to figure out whether AI mattered or didn't matter. And it feels like we're spending a lot more time talking about where it can create value and how you can deploy it responsibly. And that's starting to signal that the market is maturing.

Speaker A: One of the things that stood out to me, I agree completely in Tiffany's assessment, uh, but one of the things that stood out to me is that when we think through the consumer's lens of what is most, the most palatable uses of AI, uh, sure it is, keep me safe from fraud, uh, make sure I'm not paying fees that I could avoid, uh, optimize rewards for those customers that are in that position. But the fourth item on that list is get me to the right human at the right time. And so, you know, the. What was interesting as I think about the different institutions that were part of finovate and part of this is the design of the event. Um, the lean into AI was quite substantial. And you know, it's my responsibility to remind everybody that personal service goes along with AI in order to delight your customer base.

Speaker B: That's funny. I, um, do appreciate the practical the, or the pragmatic view, uh, of, of AI because I've had, I would say this one is the one I had the most number of conversations where people were talking about lessons learned, um, how do we actually do it and not just pilot. Right. You know, people that have done it in production and what, what um, they've learned from it and what they will do differently next time. So I appreciate those conversations. I do, um, agree. I think every conversation was AI. Um, you know what, Julie, next time I think we should say we'll give a surprise, um, reward or something for someone who can actually last through the entire demo without saying the word AI. Right. That was so much of it. Um, did, um, any of what people were talking about, did that align with your research and your data, Jennifer or Tiffany? Um, because one of the things I love a lot and I know we're going to get to it in a little bit. Tiffany, it's about your AI Visibility index. What are we doing? And are people showing up? And then Jennifer, I still remember this was back in Finn of a fall in New York when you shared with us, um, how eager consumers are trying to use the different tools to get insights and get advice. Did any of that track or not really?

Speaker A: Oh, it most definitely tracks the focus on ensuring that we have seamless experiences that work efficiently, uh, that do not create roadblocks to a customer engaging. All of that. That's at the core of what finovate demos try to communicate. All of that resonates with consumers. Um, so that is all important. And we do know since Finovate fall it's only a small increase, but we're at 53% of consumers using AI in the last three months to ask a personal finance question or for information. And you know, they're asking a multitude of different topics, uh, including shopping topics, you know, which credit card is best for me. Uh, and so I think what, what's interesting in what I saw at Finovate is that these are all very practical solutions that could have a quick impact on the consumer and that's, that's desirable because consumers are really looking for something that is a three month, nine month, 18 month impact on their financial life versus a three to five or 10 year plan. That is traditional advice that comes from banking or other financial institutions.

Speaker B: That is true. That is true. That was one year. I remember we saw a lot of um, retirement planning, longer term planning. I think this year I see less. At least I hear less is more. A lot about uh, operations, a lot about customer service, CRM, call centers, types of things. Um, so Tiffany, let's talk about the work you've been doing and I know all of us have been trying to talk you into presenting it in your analyst, um, uh, seven minute in the fall. Did you have any updates for us since the last time we talked? Because there was a lot of interest if you will, and there still is a lot of interest with regards to what institution, bank or fintech rank on top of what type of question that people have.

Speaker D: Yeah, so we run this, we'll call it AI Visibility Leadership Board every month where we go out and rank um, financial institutions using ChatGPT and we're later going to evolve it to include Gemini. But we go and we rank um, across nine different categories. And so we do that every month. And I think the biggest thing that we're seeing is that the rankings are really dynamic. It's fascinating that it's not a static leadership board and we're seeing movement month over month as bank models evolve and there's new information that interests the ecosystem in even how banks are starting to position themselves, um, competitively. So the bigger story for us that we're seeing is it isn't necessarily who ranks first. It is that AI is creating a brand new battlefield for customer acquisition. So when you start thinking about that, for decades banks spend billions of dollars competing for attention and they're now competing for recommendations. And as consumers turn to ChatGPT, Claude, Gemini or whatever platform they prefer. It really means that discovery is moving outside of those bank owned channels. So when you think about traditional search you can afford to buy visibility, right? But in AI discovery world you have to earn a recommendation. And the banks that are showing up on the leadership board aren't always the biggest banks or even the ones with the biggest advertising budgets. It really depends on how you show up in an LLM model. And it's looking at things that are very different than you, what you would look for in SEO. They're looking for things like relevance, authority, whether there's trust in that brand, whether there's expertise in that product, um, whether their reviews align closely with an associated customer need. And so when you think about SEO, banks talk about products but when you think about geo, it is talking about solutions and as Jennifer said earlier, what their actual needs are and can you give a recommendation that aligns with that need?

Speaker B: I almost feel like there's going to be a new profession, um, and a new group of people that would, that would spawn up and say instead of the SEO experts, now we're going to have geo experts and um, they're going to be running around trying to help banks, organizations or any businesses Rank you know, pop up in, in the uh, in the uh, search results. But here's the problem I had and, and I know this probably goes beyond the scope of what you're trying to talk about. Um, I tried to run the same questions with the different tools. Um, so I did Gemini, Perplexity, Cloud and ChatGPT and I had the exact same set of eight questions that I popped into each of these tools when I did back to back. So one day I ran one set and then the next day I ran the same set. I don't get consistent answers with the same questions. There were one or two questions that I asked about what's the best um, institution for freelancer. That answer seems to be more consistent. But all the other ones it's like it, it's almost felt like a random Magic 8 thing that pop up. Um, there were some that you know, like okay, yeah, this makes sense. And then the other was like who are you again? Um, so I'm not, I'm not sure. I think part of this is um, the feature, but then the other part is how do we, you know, or how can anyone actually influence a more consistent um, response?

Speaker A: I think one of the challenges here is that what you just described is no different than if we were all going to sit down at a dinner party and talk about which account, you should open up. My 20 somethings like to have conversations about high yield accounts now that they know what they are. Um, but if you were sitting down at that dinner party, your recommendations that you would get in that moment would be very different depending on who's sitting at the table. The difference here is that AI has a degree of authority to it, to some degree. Now it could be a complete fallacy that that authority actually exists, but it exists in the consumer's mind. Most consumers tell us at JD Power that they're going to verify what they learn from AI, but the definition of verification is just so broad. I mean, verification just could be, you know, asking their roommate if they think it's right. And so that's the challenge I think we're facing. I don't think the core issue of can I get the same answer from anywhere? Is different than it has ever been, which is why all those billions of dollars get spent to try to make sure the information's out there. It's just now we have this voice that has this false sense of security around it that is directing consumers in a way that may not be in their best interest.

Speaker D: And the interesting thing is when you talk about verification, you're talking about verifying financial advice from an LLM when you probably don't have the financial knowledge to even know whether it's right or wrong or to even be able to fact check it for that matter. Right. So that, that's a whole nother scary point. Like I think everyone understands that there's some hallucination that goes on, but if you don't have a financial understanding, you're not gonna know whether it is or isn't hallucinating. And then I think there's something to be said is like when you ask a question, you're giving context in the question that you're asking it. Uh, but I think there's also underlying unintentional context that is being fed from all the interactions that you've had within that LLM. So I think that is also why like for me, I, I use Chat GPT and I use Gemini, but I use Gemini less. Chat GPT knows a lot more about me than Gemini does. So, uh, if I do ask it the same question across both platforms, I, I do expect to get a very different answer because the context is not there.

Speaker B: What if I asked the same platform the same question and two different days I got two different answers?

Speaker D: Oh, that's the dynamic nature of the leader board.

Speaker B: Yeah, that was what I saw. And that's why? I'm like, oh, maybe I need to expand the sample size a little bit. It's work in progress.

Speaker D: Well, we did it 5,600 times every month. We do it 5,600 times across. So the reason. So when we do this leadership board that we have, we're doing that every month for that many times. We're also going to add Gemini to it at a certain point, so we'll have both. But that's where we're getting the consistency from in terms of the brand dimension and the number of times that it's recommended.

Speaker A: One of the things that stands out to me is that we also ask consumers, like, if awareness of personal financial management tools offered by any institution is usually somewhere between 70 to 90%, but adoption is usually between 17 and 22%. So what is it? What's the roadblock between you using these free tools that your bank that you've already selected and probably been a customer of, uh, for some while, so you have some degree of trust in. Or what is the roadblock in between that? And, you know, among the top three answers is low confidence in my own financial literacy. So if that's the case keeping you from using the trusted tools, why in goodness gracious would we just throw it out to AI and think that they're going to get it right? It's a, uh, it's a weird phenomenon in the consumer base that I just, I find fascinating because the perception.

Speaker B: Oh, go ahead, Julie.

Speaker C: No, that's okay. Yeah, I actually, I want to get your thoughts on that, Jennifer. And you too, Tiffany. I think there's a trust paradox. So when we heard that OpenAI was partnering with Plaid, and now, you know, you can get all of your finances aggregated onto ChatGPT. Um, there's a degree of trust there. Like you said, I go to ChatGPT for medical advice, for, um, you know, my meal plan each week, for relationship advice. Like, I, as a consumer have, like an emotional relationship. And I think a lot of consumers do have this quote, unquote relationship with these chatbots. And yes, I know it hallucinates. Yes, I know it's biased, but at the same time, emotionally, I have this trust there. And I think as soon as, or now that, uh, ChatGPT owns the distribution relationship with the consumer, consumers, you know, they're going to ChatGPT for financial advice, for retirement advice. They're not going to their bank. Um, so where does this leave banks? Because banks used to be the end. All be all at the end of the day. Yeah. Fintech Solve all this cool technology. But at the end of the day, banks have the trust. And now I think there's a weird relation, there's a weird dynamic with consumers trusting these LLMs m at a different level than they trust their banks. And is where does that put banks? Does anybody have any thoughts?

Speaker D: I mean, yeah, I think it puts banks, um, in the position to making sure that their content that they do have out there is not misleading. I mean, so you start thinking about LLMs, um, think about brands that operate in more than one country. Right. You ask a question about a product, and then you get perhaps details. Uh, you might be in the United States, but perhaps you get details on an offering in Canada or an offering in somewhere else. And so it will be up to the financial institutions to make sure that they're not only optimizing their internal assets to make sure that they are communicating the right information and they're being clear on what that information represents, but also making sure that they're going out to all the other third parties that LLMs use and making sure that they're optimizing and understanding the content that is out there as well.

Speaker B: Yeah. What is interesting, um, from the little test that I ran, is that some of the results that came back actually pulled information from bank websites, which means that, you know, if banks actually focus on this is going to sound really bad. But if they actually focus on being able to surface information, not just to humans to visit the websites, but also to AI agents that pull these information, then they'll have a higher chance of being in front of consumers when they're asking for certain information. Now, the flip side of that is, if they're not there, then does that mean that people will forget about them when they're looking for products and services? And if that's the case, does that mean that they start retreating more and more towards the back of the line? Whereas whoever knows how to manipulate the results in these, um, bots, they will be. They will have a higher chance of getting in front of consumers.

Speaker A: I think one of the things that underlies the trust debate is the fact that AI functions right or wrong, but AI functions today almost as egalitarian access. So when you show up and put that prompt in, um, it might know information about you, but you have accessibility to information regardless of your age, regardless of your income, your financial health status, which I posit is more important than either one of those two. Um, but your financial health status and other things, you have access to information that for many consumers in the marketplace today, banks may have excluded them from even unintentionally in the past. Uh, the idea that, you know, 2/3 of the US populace is financially unhealthy in some way, and the degree to which their bank actually personalizes content to their. To their needs today versus going after the financially healthy customer, uh, that is a current negative detractor for trust. And AI is coming at that directly.

Speaker B: So, speaking of, Jennifer, um, if we were to continue that thought. Right. What are some of the things that is top of your mind that you're worried about or indicators that we should be paying attention to? And where do you see consumers going?

Speaker A: I think we have to watch soft churn. Um, when we think about where we see consumers going, um, we have spent so much time as an industry focused on primacy and hard churn. Are we going to switch? Is somebody going to close their account? And the reality is that account fragmentation is running rampant. Um, you know, the average consumer has at least three deposit accounts, and so they have the ability to move money in a way that's easier than it's ever been. And, and we need to watch that flow of money closely. And it's possible that the new definition of primacy may be the money movement hub, like which institution is really getting the money even as it goes out. And that might not be a bad thing that some of the money is leaving the institution, which is not the way that financial institutions think today. Uh, but you being the center of how the consumer manages where their money flows is a potential new definition for primacy or is something that they should

Speaker B: have pay more attention to. Right. Because like, if I look at the bank, quote unquote, primary bank that I have, the paycheck sits comes in there, but then the minute it hits, it goes everywhere else. It goes to my kids, 529, it goes to my 401k, it goes to the bills I have to pay and everything else. Right. So. And it goes part of it to my Apple savings account because I get a higher interest. Why not? M. Right. And then the question then becomes, should you have paid attention so that you could have done more for me? So the money is not just leaving right away.

Speaker D: I also wonder if financial primacy also doesn't have a large part of it is contributed to who you trust next. So if we're saying that to your point, Jennifer, that financial relationships are fractured and you've got accounts in multiple institutions, like I would think if you have accounts in multiple institutions, then it just becomes about if you have your next financial need, who do you turn to for that next financial need? And it's who you always turn to that actually determines who you believe is your primary bank or your primary financial service provider? I won't even say bank, I'll just say provider.

Speaker C: Mm.

Speaker B: Yeah, provider. That. That's going to be the new word. Um, or would we all have a little agent that's running around trying to look for the best product for us and move money for us? I mean, I don't think that's too far. Thinking of, um, recently. Right. Nvidia, for example, they just announced, um, a new chip that they have and that's going to make all of the personal computers way more powerful, designed strictly for being able to run a local AI agent. So that is the future that they envision. That way we don't all burn up tokens, asking and uploading things to, um, an AI in the cloud. And if that's the case, would it be easier for us to move things? I don't know.

Speaker C: I don't know.

Speaker A: This is when I'm going to sound like an old woman with gray hair,

Speaker B: but I have it too.

Speaker A: But, you know, the fact that the average account holder is three deposit accounts and not seven is still saying something. Um, and the idea that we can move money quickly, uh, even today, so before all these new services are going to let us do even more and maybe do it on our behalf. Um, at some point, the human mind is unable to keep their money in 27 different places. And I'm purposely being exaggerating here. Right. Um, so I. There's a part of me that says that this will be a pendulum swing, but I don't know how long it'll take or anything in that space. But at some point, too many choices leads to a restriction of choices. And I don't know, I could see it being.

Speaker C: I could see it being fragmented. Like if a consumer had a certain percentage of their wealth or a certain, you know, um, portion that they wanted to designate to their AI each month for, you know, like Robinhood just launched their agentic Investing, for example. So not all just to kind of see how it goes. A little bit of a test scenario. I could see it. I could see that designation happening, but it makes sense. Yeah, you'd want the majority of it to be under human brain, human management, but maybe outsourcing part of it. Um, and, you know, if the a. If the agent is doing a good job, which to be seen, and to be seen also, how that'll all be regulated. And when that, when the agent Makes a mistake, how that'll all come down. But I think once that's all sorted out, I could see it swinging the other way, you know, on the, the agent side of things. But, yeah, I do agree that right now, primarily human managed is. Makes the most sense.

Speaker B: Yeah, yeah, I, I see that too. Um, I, for one so have not yet called my, uh, my, uh, the investment advisor, I think that's what he's called himself for like, I don't even know how many years now. Ever since my money is just goes to the 401k. Um, he's like, theo, we need to chat. I'm like, yeah, no, I don't want to talk to you because I don't know you. But would I trust an agent more in the future? I don't know. Or maybe I just don't trust any of those. I have no idea. I don't even trust myself, clearly. My kids told me I made poor choices, so. But, you know, with that being said though, what. What should we look forward to, um, for the rest of the year? I. I feel like I can't keep up. Right. So I was, for a very short period of time, I was watching, um, Open AI with Intuit and then Intuit with Claude. I'm like, oh, what is going on there? Next thing you know, now we're moving on to OpenAI with, um, Platt. And then there is Robinhood. Julie, to your point, I'm losing track of exactly what's going on. I have a feeling by the time this episode launches, there will be like five different headlines out there already. So, um, what are you watching, Julie?

Speaker C: So, yeah, I mean, I'm certainly watching similar partnerships to OpenAI, PLAID. So shifting distribution channels is one thing I've had my eye on. I think it's going to change right now. Um, OpenAI's partnership with Plaid, it's available as read only, so no agent involvement. They're not going to be able to shift funds around within there. Um, and it's also only available for pro customers. So that's, you know, at the 200amonth level, I think. Um, but currently, I think the current stat is that 200 million ChatGPT users, um, are already asking financial questions daily. And so with that type of demand, you expect to see, um, similar types of partnerships at different LLMs. So I'm curious to see what that will look like, if it will change from read only, um, to, you know, having capabilities like Robinhood launched. Um, limited capabilities, because, again, the regulation isn't there yet. Um, but, yeah, Limited capabilities for agentic payments, agentic investing and things like that. And I am also theo very behind on the Agentix space because I feel like it's every other day or certainly almost every day I hear of a new Agentic Payments launch in my inbox this morning was Q2's now um, in on the agentic game for banks helping them automate things. So it's very hard to also to figure out what's real and what's not. And um, you know I think unless you're in it every day and you're actually using the solutions, it can be hard to figure out what's real, what's really moving the mark and what is more just marketing. So that's what I'll be spending my time doing, figuring out marketing versus uh, what is actually going to make a difference here.

Speaker B: Wow. What about you Jennifer?

Speaker A: So I think I'm going to have a little bit more of a short term focus at the moment which is investments have been made in those initial things. When we first kicked off our call we started talking about AI for customer service. Um, in our data I'm going to be watching whether or not problem resolution actually becomes easier to achieve, whether or not customer satisfaction with the ability to resolve friction. Um, and let's even move beyond that moment and think about is it easier to open a new account, is it easier to get financial advice and advice and guidance. Three of the key moments that matter that could be influenced by the different investments that have been made in near term customer experience. I'm going to be watching to see whether those actually move because if they fail to move that will have long term impact on trust of agents as they emerge in other ways. Uh, and so I think that's the short term thing to monitor.

Speaker B: And what about you Tiffany?

Speaker D: I'm going to go along with uh, Julie here. I um, think the most interesting thing that's happening right now is really seeing AI platforms move from being treated as technology to really understanding that it's a new distribution channel and that today it's really focused on helping people discover products but in the future it's going to focus on helping them compare options, to make decisions and to ultimately take action. So I know we've talked about the Plaid example, um, but NatWest also launched an experience inside of ChatGPT for prospective home buyers. And I know the three of you know my love for homeownership and the journeys around that, but imagine if you could, if you could actually um, help a consumer with home buying within ChatGPT and go through the education. So I think what we're going to start to see over time is again I said this earlier, we're going to start to see this leaderboard continue to shift, uh, as brands start to understand how to be recommended in LLMs. But I think we're also going to start to see banks try to figure out ways to insert themselves more clearly without having to rely on that Recommendation engine, as NatWest has done.

Speaker B: Yeah, then I think mine will be a combination of all of those. I'm very intrigued with how technology is evolving to the sense that what are the different players doing? To your point, Tiffany, what are the banks doing? How are they coming back? Um, and to your point, Julie, what are the um, Frontier AI providers doing? What is Cloud doing versus what is OpenAI doing? And how all the rows shifting? Um, how much do consumers trust them? And you know, for the last two years I think we've seen a divergence of trust in the sense that certain demographics, in certain countries, their consumers seem to trust these tools way more than the other ones. Right. And, and that gap seems to continue to grow. I hadn't seen them close yet. So I think that is absolutely intriguing. Is, are other people trusting, trusting these tools more because they feel that they're more accurate, or is it because they feel that there is accountability and there's repercussions if things don't work versus, you know, could the regulation or lack thereof that we have here create this little conundrum that can we trust it? Maybe some and maybe not just yet. So I think there's still a lot, lot of song and dance that's happening in the space, but never a dull moment, I think. That's what I would say. Um, and until then, I look forward to seeing you all in New York, hopefully. And um, let's see if there will be more interesting topics for us to talk about beyond AI. AI is interesting, but I would love to see it being used to help small businesses, to help people actually close the um, affordability gap a little bit more. And um, but thank you all of you and, um, appreciate you and this is a lovely way to um, get ready for summer. So until next time, thank you for joining us and thank you very much for joining us for another episode of One Vision. We'll talk to you next week.