9:21the session, we've structured it this way. So we're gonna cover, kinda, like, the four areas, where we felt that we have the most, the most value to provide to you guys. So first of all, we're gonna do a little bit of a reality check. So why generate, why generic AI content coming from, let's say, generic off the shelf generative AI services aren't necessarily delivering the speed improvements,
9:47the banks have expected or that have they have seen maybe in other industries, and where the actual bottleneck lies. Second, we're gonna go into the behavioral intelligence. Ultimately, we're talking about marketing. Marketing is the science of using behavior to drive decisions. And so what a decade of campaign data across financial services reveals about the language patterns that drive real outcomes,
10:15funded loans, card activations, account openings, and why generic AI consistently misses them. This is something that that's this this will be Lisa's spotlight moment. She's gonna walk us through a specific case study that I think will challenge how you think about content testing. And we're gonna talk about, obviously, the elephant in the room, something that is, you know,
10:41in in in the air and in the water and in the ventilation of of the financial industry, which is compliance. So we'll talk about what it actually means to build compliance into AI content generation rather than just, like, bolting it on at the very end. And we'll share something about the terms of services. Now maybe that every major AI tool that most banks might not even be aware of or they might be aware of, and that's what's putting them off from using them.
11:09And fourth is the human layer. So in a world where AI generates the content, what does the human expert actually do? What are what are their jobs? Because the answer the answer is not nothing. It's just something completely different from what it used to be. And as I said, we'll save about ten minutes in the end for questions. Okay. So let's start with Lisa.
11:35And so, Lisa, most, banks, credit card unions, fintech, financial institution, businesses in general, in this audience may have, experimented they probably have experimented with AI to some extent. Chatgpt, Gemini, Claude, maybe more specialized tools around creating content for marketing. But, really, whatever's available and whatever's being made available
12:02and and been whitelisted for them. What are you hearing from institutions, from from from customers that we're working with about what's actually happening with that content once it's generated? What's actually happening is that we're creating more content than we can possibly use. We are creating so much content today.
12:28We've removed the barrier to content creation with AI, which is amazing. It's like we removed the barrier to distribution of content previously with the Internet. We're seeing a huge shift in our society, and by removing this barrier, we are flooded with content. This leads to two issues. One, which I think we'll get into later, is, well,
12:56but what is good content? Is all of this good content? And do you have the data to say that it's good? What does good mean to you? To me, that's intelligent content, which is where I work. But then there's this other problem, which is having so much content isn't really great if you can't get that content into market. And that's what we hear a lot from the companies that we work
13:22with, is they have so much content, but their workflows with the different departments in their organizations to take the content and prepare it and execute sending it, and more importantly, get it approved to be sent. There's a lot on the line anytime a financial institution sends something out into the world. And making sure that they are safe,
13:50that this is good for the brand and good for the consumer, that takes a lot of time. So just having infinite content, I mean, it's a bottleneck in a way. It's worse. Yeah. I I think I think I think I couldn't agree more. This is also something, you know, harkening to something I've I've I've said earlier. I think it definitely surfaces, especially in highly regulated industries where every letter counts,
14:17and the list of things that you can't say is ten times longer than the things that you can, and you still need to operate within those guardrails to create a good, successful campaign that performs, that can be measured, that can be iterated on, that that problem has not essentially been solved by by by the the the ubiquity of generative AI even if it, you know,
14:43even if it is used for initial experimentations and for for ideation for the creation of the brief. So a soft question for you. So you've been working with the largest banks in the world for over a decade. You've seen the the evolution and and and how their behavior and approach towards this has transitioned throughout that period of time and more specifically
15:10throughout, let's say, the last three or four years. When you look at the broader market in approaching AI content right now, what is the most common mistake that you see in these, really household names, of of of banks and credit card, mortgage companies, and so on. So if we go to the previous slide for a second, I'll explain.
15:36So in general, as as we said, AI is is a way for, at least conceptually, to augment and speed processes and outcomes. But what we've seen over the years is that companies, specifically banks, struggle with that and struggle with the ability to actually change a process
16:02that usually takes four to eight weeks, actually, sometimes twelve weeks, to get from an idea or a content to market. Now to just to put it in perspective, if you have a change of rate, for example, by the Fed, and you want to change your communications around mortgages or loans, and you are six to eight weeks late, just that actually reduces by fifteen to
16:32twenty percent your conversions from any campaigns that you're doing because you're missing completely on the early switchers. Given that we work with some of the more advanced and, I would say, competitive banks out there, they have realized that. And what they are trying to do is not look at AI as something that allows me to get from an idea to a draft, but actually to look at the entire process and say,
16:59how do I get to market faster by not only applying AI to create fast, but also accompany the decisioning that is required to get to market. And this decisioning has two areas. One is is what I'm doing the best it can be from a performance or objectives perspective? So I need cards. I need funded loans.
17:26I need funded accounts or or more deposits. That's one aspect. And how do I justify that? And the second thing is, of course, whatever I'm doing needs to be regulated or is regulated and complex. So suddenly, I I need to compete with more companies out there. So as many of you feel, I would say, day to day,
17:52there are a lot of options out there both from fintechs and established banks, retail, other community banks, or or credit unions. I need to be fast. I need to perform, but I also obviously need to be safe, compliant wise. How do I condense all that to something that will give me a competitive advantage? If you remember something from that part is how do I
18:20turn any marketing asset or communication to become a financial asset such that it has an impact on the business? Okay? So that's what we've learned. And we've seen that, as Jonathan said, it's not about ability to create a subject line or an email body or a social post or it's about the entire process of getting that to market.
18:47And there is another part, which is once it's in market, how do I learn from its behavior or the behavior it it drives for my next creation or generation or writing for that matter? How do I do that in a way that allows me to compound the learnings? This is what some of the largest have
19:15struggled with and, again, some of the more advanced ones have focused on. And, in our case specifically, drove us to create the capabilities to to enable that. But in general, if you remember something from this slide is you need to think about the workflow and how you condense that because speed of decision creation
19:42and regulatory compliance allows you to significantly better compete in market. This is the essence. Yeah. Those those those those words hit hard, I'm sure, for a lot of a lot of our audience members. So, yeah, I think something something to take away from this is is, again, that the true and core
20:09value of a campaign doesn't lie with the ability to create it and the ability to write the brief in seconds, but it is to get it out as as quickly as you were able to create it while ticking all of all of the boxes, while remaining both compliant and and and and performing, and keeping a very,
20:35very strong grasp on your ability to learn. I think a lot in a lot of cases, specifically with with generic generative AI tools, there is almost no learning, let's say, beyond the the the the single session, definitely not at the campaign level. Okay. So so as we said, ten percent has has has always been about the execution and the creation of the content.
21:02Ninety percent still remains the the heavy lifting that still needs to be to be handled and that most off the shelf products really can't can't address properly. Okay. So moving on to talking a little bit about the data that we that we have been analyzing and that has really helped to shape a lot of
21:28our perception about what what performs at the intersection of a good campaign and a compliant campaign for financial institutions. So, Lisa, you've analyzed hundreds of thousands of campaigns across the largest financial institutions. So when you look at the language that actually drives someone to open an account or fund a loan or activate a card
21:56or choose a mortgage provider, so what does generic AI constantly get wrong with that? I think that the issue that generic AI gets wrong and that a lot of humans get wrong if they're not digging into the data and measuring That's like opens, clicks, conversions,
22:33sign ups, engagement on some of the social channels. And what I've learned over the over the time that I've, you know, been working in this area is that it's not just the simple campaign details that are going to drive that. You need a lot more. So you need to really and this was the surprising one for me.
23:00You need to have emotion. You need to have language that really captures the audience, and it may not be directly connected to that offer or to that product, but it's going to connect with the person because there's a person behind there who is seeing, reading, hearing that message. And so when you do that, you get a much better response.
23:26And this I've seen it apply across different financial services products, across different channels, and across all sorts of different brands and organizations. You always wanna have your content make sense and be relevant to the campaign, But you need to take in, this sort of data from testing and learning in market. As Asaf said, we really need to measure and test and take the data back in
23:54in order to learn that these things, like emotion, work. And then, of course, which one? There are so many emotions out there. How do you tap into your audience? So for me, I think that where we get content that converts is building an AI that's continually learning through this testing and seeing trends like this, some of which are counterintuitive,
24:22and funneling that always back into the content that is created, and, of course, making sure all of that content that is created is on brand and is safe, is compliant, and is gonna work for you. So, I mean, let's let's try and make this, and make this concrete and put this, into actual examples. So this is something that we've, that we pulled off from,
24:49one one very successful campaign, that we ran for for a a Fortune five hundred credit card issuer. Can you walk us through kind of, like, what was, you know, the the differences between the control and, and what we applied? Yes. So it looks like we only have the headline on the screen, but the headline is one of the strongest touch points. It's usually very attention grabbing, and although we don't have the creative on the screen or the
25:16formatting, usually it's big, it's loud, and so it has a lot of impact. And the conventional approach is leading with a really big number, which can be eye catching. We have five fifty thousand loyalty rewards points. But it's leading with the specifics, with the offer, and with those offer details. And this might be, probably is, a really enticing offer, but it's also really factual,
25:42and it's not really resonating with the audience who's reading that message. So instead, over on the right hand side of the screen, we've switched to the emotion of gratification using the wording you'll get. You getting this is exciting. That's like Christmas. You're getting something. That automatically gets more response. Starting with you instead of a, you know,
26:07command like get, maybe instead of an imperative verb, starting with I won't get technical. That's me being a nerd. But in starting with the words you, okay, people respond to you. That's a great way to get started. And then we've also added a different value proposition here, a reason to believe, so to speak. Instead of talking to just the points themselves, we're talking to the idea of convenience,
26:33something that people really want, especially in today's fast paced world. So we're starting with emotion, then we're leading into convenience. And here, you can imagine these headlines are part of the same, let's say, email or web banner or something. And there's a whole long message that talks about this, you know, speed, sixty seconds, and also talks about all the points and has a lot of different We're trying to pull the things that people are
27:00going to emotionally respond to here in the headline. Yeah. Yeah. I think I think it's it's it's really important to remember, you know, as we automate more and more areas of our life and we offshore them to to AI, that, ultimately, these are humans making human decisions, and humans need to to be talked to and and to be they need to feel something towards whatever it is that
27:28they're being sold. And and so the we we can never undersell and and and underappreciate the importance of of even just, like, a word or a verb that will that that will remain unchanged so long as humans are humans. Alright. Asaf, back to you. Alright. Let's, you know, let's let's let's bring up the elephant in the room.
27:54And so, like, every compliance officer in a financial institution has the same reaction to AI generated content, which is how do I know this isn't going to get me into trouble? How do I know that, you know, whatever Claude or ChatGPT spewed out is you know, it might be true to the prompt that they were given, but, you know, banks are under different types of scrutiny
28:22than than your average marketer in any other business. What what what can we tell them? Oh, you're on mute. I'm muted. Yes. Let me start at a higher level. First of all, many community banks and credit unions we talked to, they first of all, they they have some struggles in getting AI into the company because of governance and model governance processes.
28:51I just want to say on that front for those who suffer from that, marketeers who want AI is you should fight for it because done right, it has a lot of value. So just just a a word of of of of empathy to the process of getting AI into the company. Secondly, in general, it's not just about
29:16AI that compliance officers are worried about. It's just content and whether it puts the company at risk or not or creates some kind of liability in that front on that front. Now in general, the process is I'm creating something, and I'm sending it to compliance. And then I have some back and forth, and then, hopefully, I get it approved. And sometimes it feels subjective,
29:43and I'm not sure why it's not approved. And sometimes I'm just deciding not even to try because I know that the process will be too long for me to compete in market. Now this needs to change. The issue with AI to draft content is that it doesn't solve for that because it remains the same. At the end, the compliance officer gets something and
30:09reduces any any any issues by eliminating ways you could and wanted to say things. And then it goes back and it goes to market with less impact and less less conversion, less less business value. This is when you apply compliance as a bolted on capability, whether it's with some AI tool or just with a human. Now what you really need to do,
30:36and we we we realize that quite lately, is that you need to somehow generate content that will have all the reasoning within it of why it is compliant. And this is somewhat hard to do because you need to generate the content, have that content be performing, but also compliant. Now today, it's either or.
31:04It's like a one handed clap. To really have an impact, you need both functions to work at the same time at inception. And this inception is I'm creating content that is compliant, has all the reasoning of why it is, and therefore, ninety, ninety five percent of the time, it will get approved as is. Okay? In that context, I'm not only able to go faster to
31:31market, but I'm suddenly able to personalize. Because personalization, if you think about it, is essentially having multiple experiences, and these multiple experiences even for a credit card or a checking account or a mortgage or a loan, is taking an offer and creating multiple experiences that are then delivered to each person based on the one
31:59that fits best. Personalization means, in general, even more content. And most financial institutions struggle heavily with that while wanting to personalize. The only way to personalize is if you compress that process to minutes or hours, days, if you can if you if you cannot. And if you do, you get both the benefit of
32:28time to market and the benefit of personalization. So that's something that is extremely important. But, again, to answer Jonathan's question, compliance cannot be something that comes after if you want to compete properly and get disadvantages. It needs to be together. Otherwise, doesn't matter who created the content, it will get reviewed,
32:53and you will still have the back and forth. And this back and forth is what drives most of this time to get to market. Okay? Yeah. Absolutely. I think, ultimately, it it it it really is down to which which areas, of of the process, are being are being eliminate eliminated or or accelerated and which still need, you know,
33:21need quite a quite a bit of scrutiny. And I think, again, this is a this is kind of a it's it's kind of a side note, and I think this may be is a is a callback to something Asaf opened his response with of, you know, those, you know, those of you who are, I guess, struggling with with with working generative AI tools or maybe who you you have worked
33:48and this is something that you are trying to offshore a lot of your compliance a lot of your content creation into is that a lot of these solutions really kind of disclaim and try to distance themselves from from providing and and being liable for compliant content. So while that is very, is a very important and effective tool for the initial stages,
34:16when it really comes to content that will not only look good or fit the brief but will also keep you out of trouble, there will always have to be that extra layer. Lisa, back to you as the as as the human ambassador of AI for Sato.
34:41Just coined it. So we talked about what AI can miss. And so let's let's double down then. Let's talk about the human side. So in a world where the content itself is essentially most of it is is AI generated, and there's a huge influx of content, more than you know what to do with it, what does the human expert actually do?
35:06How does their job look like? There are two types of human experts working in AI today in my mind. One is the human expert who is building with AI specialized solutions. That's like my team at Persado. We are human experts. We are working with AI. We love working with AI, but we are incorporating our expertise so that we have a
35:34solution that doesn't have the type of risks that Jonathan just showed on the last slide. We don't want I mean, those solutions are trying to do it all, absolutely everything. So they're not going to have expertise in absolutely everything. And so bringing in humans who have some layer of distinct expertise to help train and validate and test, that's one type of human
36:01working with AI today, and it's an industry that's booming. There's another type of human, though, an expert human, and that's the end user expert human of the AI. When you go and ask one of these models a question in an area you don't know very much about, You don't know if the answer you got is correct. You can't be one hundred percent sure. It probably sounds logical, and you might decide, yeah,
36:29I'm gonna follow that and see it through. But if you're an educated user, if you are an expert marketer, because that's what your career is in, if you've been working on a financial services brand and you know the industry and you know your brand and you know how to write, you're an expert user who can be a critical judge of what outputs you receive from AI and can help tune what you're gonna get in the future.
36:57But your role has changed even more as a marketer, perhaps, from a writer. You're a critic. You're a reviewer. You're also a strategist. And I will say that as AI changes and agentic systems become more and more powerful, which is happening, maybe the role of the strategist change changes, and maybe the AI will do more of the strategy.
37:22I do think it will. I think its role will expand. But even where we are today at this moment in time, the human layer is a lot of strategy, is a lot of bringing knowledge and directing AI to add value in places where we as humans know that other humans who are consuming products from our institutions will find value. So there's a lot of human human component here that you,
37:49as an expert, are a big part of. And so I think that your role has changed into one that's a lot more critical thinking. But to me, that's really exciting, because, of course, those are the pieces of my work that I like to do the most. So I think that, in a way, a lot of opportunity has opened up for the person who is a user of AI in any way, shape, or form.
38:16Yeah. This this this reminds me of of something I read in a in a recent report published by one of the consulting firms about how how companies might handle the junior apocalypse that is upon us where you have, you know, people that are fresh out of universities with zero practical skills that can very, very easily be replaced and are already being replaced de facto.
38:45What do what what do we do with these people? How do we get them experienced? How do we how, you know, how do we make them function in the workplace where a lot of a lot of those entry level skills have already been made almost entirely obsolete by AI? And one of the suggestions, and it's not it's not without its challenges and and hurdles
39:11is to really kind of, like, give them a crash course at the business reasoning of the company that they work in and have them train as these sort of AI liaisons. Because as you said, they you know, the the AI spews out something that looks good, that reads professional, but you, as someone who maybe is is new to the field,
39:36won't really know whether or not that was the right thing or that will be what will work, what will be impactful, what will sell. And so one of, you know, a a potential future job that this, you know, industrial revolution two point o that we're that we're living through will will help create is these, like, agent liaisons, people who will be brought in very,
40:02very quickly into subject matter expertise and will help translate output into goals. Am I agree a hundred percent. And I also think that deep subject matter expertise, even people with very little experience in the workforce who are coming right out of universities, can come with very deep subject expertise that can be very
40:28valuable in a field and can quickly learn some of these other skills. I tell everybody I'm a linguist, and I am. That's what my degree is in. But my subject matter expertise is in automatics, which trivia for some of you is an expert in names. And so I, at twenty two years old, brought an expertise in a deep in subject matter that a lot of people did not have.
40:53And I think so many young people joining the market bring deep expertise in a thing that they are passionate about. Yeah. And and it's it's always important to remember that it's the softer skills that are that usually, you know, kinda permeate most effectively. So from a different angle, Asaf, you've built a company essentially around this hybrid
41:21model for over a decade, this sort of AI agency, if you will. And what's the biggest thing that that had taught you about where and why keeping humans in the loop as opposed to where actually they shouldn't be? And you're on mute before I end. As Lisa said wonderfully,
41:48for AI to be impactful, it needs to be somewhat verticalized and profound. For that, you need expertise. You need expertise both in this context, financial services, their processes, their limitations, their regulations, and to be able to even, in our context, train AI to understand these subtleties about
42:16this context. People on the call that struggle with UDAP know how subjective that is. You need really, really deep expertise to translate that seeming subjective matter to something that can be implemented with AI. That's one aspect. The other aspect is to understand the process and fit it in such that it has
42:43an impact requires humans today. Okay? That's another layer of a consultative approach to implementing AI. MIT published a now very cited publications. Ninety plus percent of AI implementations fail. And they fail because they are not fit for purpose. That's the reality of it. I'm doing some prototype.
43:08I'm doing some trials, but I'm actually not solving for the core issue, whether it's a process or other capabilities. For that, you need humans. You need humans to analyze. You need humans to understand, and it needs to fit. The other part is and then where you need humans is we hear a lot when we go to a finance institution. Say, okay. You have this AI. You can do things.
43:35I want it to fit in my process. So I'm using Excel. I'm using emails. I'm using the system. I'm using that, and I don't want to change anything. I just want somehow AI to fit in. This means that you get a fraction, maybe ten percent of the value. If you're not thinking along with humans that understand your process, how to change it to make the most out of that
44:04capability, then you just won't. So this is where we where I see this combination of human and AI. At least on our side, there is the aspect on the, obviously, customer side and, as Lisa said, the expertise of understanding that AI can do things and how do I judge that. And then going to the next level, which is, okay. I don't need to write,
44:29and I don't need to evaluate whether it will perform. It tells me what are then the objectives I'm setting for it. And how do I connect it to the business, which is a shift that needs to happen, but also a shift that is happening from humans. And the last thing I would say and connects to something I've I've thought of saying before is when we say
44:55AI, there are two elements to it. One, me as a credit union or community bank or fintech for that matter, how do I use AI to to improve my processes and capabilities and impact? There is the other part which sometimes people forget, which is how do my customers use AI in this new world? So about there is a study from Mintel.
45:20Some of you know Mintel. I think over sixty percent of consumers have used AI to ask about financial services product. Most of them trust the ChadGBTs of the world more than their bank to answer, should I take this credit card or the other credit card? But the decision is still the human. And there is a conundrum now where you need to be able not
45:48only to talk to humans, but also to somehow convince machines to cite you, to make you a reference. So just just a broader thought about AI, which is as we move forward from now to the next year, two years, there will be more and more evaluation of how do I communicate the right answers to those machines, but how do I connect
46:15emotionally with humans to make the right decisions? This is critical because a machine can only give you the rational argumentation, but these are actually not the drivers for humans to make decisions. So you need to do both. So, anyway, it's a long way to answer. Human plus machines as a service as well is actually very important for the adoption and scale of
46:43AI within a company. We haven't seen AI scale if it's not accompanied by humans to help you absorb, connect, adjust the process, and then really scale to make it truly impactful such that you can go back to your CEO and say, we've actually changed trajectory, which is rare today if you just use
47:10a generic model. And, yeah, I I think the the the the bottom line here is, like, once once we've once we've outsourced the the the tedium and the drudgery and the parts of the job that we have always disliked, what that leaves is it really heightens and and and enhances and emphasizes the human element.
47:38And nowhere is it more is it more palpable and more measurable than than than in marketing, than in trying to use, you know, human emotion and behavior to drive a specific action while being surrounded by millions of data points to to tell you whether or not you're doing the right thing. Okay. We are at the last ten minute mark,
48:07so feel free to start submitting your questions. Let me take something that or is already up there. Okay. So here's one. What is the right way to get a compliance team comfortable with AI generated content? Is it about the tool? Is it about the process? Is it both something else?
48:34Lasaf, you wanna take it? Sure. First of all, they have to see it in action. I can say that on our side so we we we've been generating content for almost ten years. We've been rejected by compliance officers more than anybody else in the market. And to make them feel comfortable, they need to see that the pattern of their decision
49:01making matches what whatever AI is doing on that front. It's not only to point out what's not compliant, but also provide the reasoning for that. Once they see that, what usually happens is that you have a phase one where you reduce the back and forth between marketing compliance by about ninety percent. So ninety percent of the times,
49:28you'll get just first round approvals. That's very important. And then the phase two is many of our customers want to get to a place where they have a traffic light system. So they trust the AI after they've seen it across different communications work well, and they now want to move to a place along with the marketing team where anything that is flagged as green, so compliant,
49:57can go to market as is along with the audit trail of what was analyzed and and why it is compliant. As a company that trains agents to do that, we, on our side, we have and this is Lisa's team's responsibility. We have a grounding set who evaluates across hundreds of communications annotated by experts.
50:22So a legal team of legal people that know a bunch of stuff that I have no clue about across the twenty plus regulations that apply. And they annotate and say, okay. This is wrong because of this, and this is wrong because of that. And then you you compare that to what the AI system produces, and you make sure that you have, in at least in our case,
50:48about ninety percent accuracy and ninety percent recall. So you catch most of the issues that the compliance officer would catch with a high level of accuracy. But, again, you start. They get they get comfortable. Sometimes you adjust based on the company's legal guidelines, which happens usually at the start. But the AI system also learns from comments,
51:13and the compliance officer learns that the system is good enough to provide substantial feedback or valuable Alright. Great. So on a on a on a similar vein, Lisa, see if you wanna you wanna take this one. How do you measure whether AI content, actually performs better, versus just being produced faster?
51:42We set KPIs to measure what performs better. We say that better we decide what better means. Does better mean more clicks, more sign ups, more applications. Maybe it just means engagement on a social channel. What is more? And then we measure that. We run tests. Persado is known for going way beyond the AB
52:08test into sometimes I like to think of it as, like, a through k tests, using an experimental design method. We're able to test into the smallest pieces of language. But we run a lot of tests, and we collect that data. And that's how we know what works. Then we use intelligent humans and also AI, and machine learning, together to analyze patterns in response
52:35data to our KPIs and to figure out what it is that makes people take that action. That's how I can say emotion works, because we've analyzed it and we know. And if I know more about your brand or your product or your channel, might be telling you achievement works. Speak to rewarding people. Say you've earned something. That might be a motivator,
53:03an emotion that's going to drive action. But for somebody else, for a different campaign, maybe it's safety. Make me feel like it's true. Seriously, this is real, confirmed. If you use those words in the right context, again, not overpromising, you're able to set the right emotion to the right moment in time, to the right campaign based on data that we tested
53:30and knows know we know what works. Let let me complement that just to just to give you a sense. Lisa mentions emotions in language. About seventy percent of the impact of these communications across channels except social, I'll say something about social in a second, is driven by emotional language and specifically by language. So just to give you a sense,
53:56changing the language in an email, for example, versus changing the image, the language is way more important to drive impact. And within that, the emotional language. Okay? That's what it means. And, again, it has to do with you communicating with humans on the other side. I see another question here, Jonathan. I'm not sure if Yeah. Yeah. I was gonna but to to to get to that. So, yeah, last question.
54:23So this is for yourself. Okay. Earlier, you mentioned how marketing communication should be a financial asset, and I love it. But can you elaborate on how you said something about process needs to be in place? So this is specifically I think you you made a mention about Yeah. Responding to a rate change, and I think this might be something about how to,
54:49say, using AI or applying AI to to a workflow when when such a fast reaction time is needed so we can quickly respond to that. Alright. So first of all, taking a marketing asset, making a financial asset means that when I talk to customers, it's the best way I can talk to them. It's in markets in the best time and therefore
55:16drives immediate financial value to the institution. It's not it's not about using AI to design the right workflow. This is actually something that should be done between humans and humans, so the team and the bank and, in our case, Persado. But it is about using AI to take an idea. It can be a brief. It can be a current asset.
55:42And providing the new creative, new communication with all the intelligence that says, why is this the best way to communicate for that on that channel, for that segment, on that product and in a compliant way. This actually changes the time to market. On top of that, there is how does it fit to a process.
56:07If I have a workflow system, can I initiate it from my workflow system? How do I do the submission to compliance and back and forth? How do I move this content from whoever created it to the distribution channels? It can be an ESP or the web or mobile. This is on the process side. If you want to have the full impact, you need to have great content produced fast and compliant
56:33with all the performance capabilities and put it in market in a way that allows you to communicate fast in terms of distribution, but also get back the results such that the AI learns from that for the next iteration. Now there is another question here around it seems a lot of value from AI is based on speed, but shipping faster does not mean we're shipping better.
57:00And this is correct. If I wanted just to ship faster, I could maybe take out some of the compliance restrictions, and I create something. And this either means that I'm not safe in my communications from a regulatory perspective or that I created something fast, but it's not performing in market. The right way to do it is to take in this
57:25behavioral data and know that when I communicate something, it's not just faster that has a lot of value on its own, but it is again the best way to communicate to that type of customer or segment or audience. It's the combination that matters. Doing things just fast, even if you could from a process perspective, is not enough to change the impact. If you do that correctly,
57:51I'll just give you a sense you on average, you get about forty three percent more conversions if you apply the right content with this intelligence, with emotional language, with narratives. And if you do it fast, you get the benefits, of course, of what I called before the early switchers. So yeah. Okay. So I'm, yeah, I'm just about to ring the bell,
58:18which segues us I know. I know. Really, really nicely, I think, into, you know, the the five points that we can partake to you and when you think about how AI looks for you, for, you know, content in the organizations that that you work in, wherever you are in the adoption process, whether you are
58:44neck deep or you're still kind of testing the water somewhat apprehensively. I think it's it's important to remember, I guess, these kind of five points. So, like, one is that faster draft not mean a faster campaign. I think that's exactly what Asaf has just just talked about and something that we ran through different angles of. The ten ninety rule, so invest in expertise,
59:10not just execution. Ex execution will always be the tiny little bit at the beginning and not the actual the actual value of the campaign. Generic AI misses the behavioral patterns that drive financial decisions. For that, humans will still, hopefully, for the foreseeable future, triumph. Compliance during generation not after is the only way to
59:38not compromise on on safety and on liability when when we're operating in this generative content world. And last but not least, build compounding intelligence, not disposable content. I think it's it's that that keeping your content within a system that learns from that content being in market
60:03and applies that learning automatically or semi automatically onto the next campaign and the end campaign after that will will ensure that all of that time did not really go to waste. So with two minutes over, I wanna thank Lisa and Asaf for your time and your and your wisdom and experience.
60:30I think I've this has been really fascinating for me to to observe and to moderate, and I hope it's been that for for our guests as well. Thank you for the financial brand for having us, and we wish you all a very, very good rest of your day. Thank you. Thanks, Jonathan. Thanks, everybody.
