Leveraging Generative AI to Write Stuff!

One of the most common use cases for utilizing AI is to assist in writing and creating content. Join us for an insightful session on how Generative AI can revolutionize your content creation process. Discover how the various Large Language Models can help you craft compelling emails for clients and prospects, enhance your website messaging, and generate engaging blogs, articles, and newsletters. By incorporating this amazing technology, you can save countless hours and redirect your focus to other revenue-generating activities.

Our panel of experts will discuss best practices, evaluate the most effective tools, and provide live demonstrations. Learn how to seamlessly integrate Generative AI into your workflows and start reaping the benefits immediately.


Transcription:

Diana Cabrices (00:10):

Hello everyone, and welcome to today's panel. If you're here, you are here to learn about AI and marketing, and we are so very excited to talk about this topic because we use it in our own businesses every single day. So I'm just going to quickly shout out this amazing star panel here. I'll start with myself. My name's Diana Cabrices, says, some of you might've seen me earlier during the Technology Demo showcase, and I'm joined here by Tim, Mary Kate, and Susan, who all work in their own respective businesses in this space. Tim's a Financial Advisor, Mary Kate and Susan are both Marketing Executives running very successful companies in the space, supporting financial advisors and FinTech firms. So we're going to get started, and the first thing that I want to do, and you'll notice throughout today's panel is that as we talk about a specific question, you're going to see it up here on the screen. I want to make sure everyone is tracking. Now, this topic is timely, it's important, it's obviously very relevant to the conference that we're at today. So I thought we'd start first by just defining what generative AI is, what large language models are, and just breaking that down into simple terms. So from the advisor's mouth himself, Tim, I'd like to hear your take on this.

Timothy Bickmore (01:30):

Okay, it's a big question to answer. I think if people know about large language models, but coming from math, Wisconsin, obviously we have a lot of corn fields, so I'm going to use a corn maze as our analogy to simplify what a large language model is. So if you look at a corn maze and it's the size of the United States, imagine that's your matrix with a large language model, and you have to figure your way. You put someone in the corn maze and they have to figure their way through the maze, right? There's dead ends. There may be false starts. And so the best way to kind of explain it is imagine that you need directions. You have to give directions to this person that's prompting. So if you don't give them great prompting, meaning, hey, maybe it's a left up here or a right there, that's not going to get a very good outcome.

(02:16):

But if you give them specific prompts, meaning take two steps, then take a left, then take three steps and take a right, then you can really start to get better, more efficient outcomes within your model. Then if you imagine giving them a compass, right, that's more data to be able to get through the maze more effectively and more efficiently. That's your data quality. So if you give them a broken compass, not really going to help. But if you give them a really high tech compass, then they're probably going to get through that corn maze pretty well. So essentially that's what the large language model is, is that you're trying to get through this corn maze, but we've trained that individual a thousands and thousands and thousands of times, but you still need to make sure that you're prompting it, giving it good directions, and giving it quality data to get the output that you're looking for. So when you're tapping in ChatGPT, imagine that little person running through that maze to then try to give you your output back, essentially to simplify it. That's what a large language model is.

Mary Kate Gulick (03:12):

I like your metaphor, Tim. I like it better than mine. I am such a literal person. I mean, LLMs are just the brains behind most of the AI tools that you use. And they're trained. They're large language models because they're trained on just mass amounts of verbal data, books and articles and video and audio. And because of that, they're really great at verbal tasks. They're great at summarizing data and text analysis and writing code. But there are things that LLMs are not great at. It's only one type of AI and LLMs are lousy at math. While they're good at certain types of data analysis, there are words, language is imprecise, and that's why you can get things like hallucinations in your outputs. So understanding that large language models are trained on human language. They are as imperfect as human language is and what they've been trained on. And generative AI, it's exactly what it sounds like, and I think you all know the definition. It's using that type of artificial intelligence to generate new content that never existed before, which has been a boon to content creators and also a source of frustration to people who can't necessarily get it exactly the way they want it.

Diana Cabrices (04:28):

Absolutely. So we've got corn, mazes, hallucinations. This is going to be a very fun session. So thank you both for weighing in on that. Now, one of the things I want to cover early ahead, if you read the description for this session, you'll know at the end of our session, we're actually going to show you some of this stuff in action, how we as marketers on this panel use AI large language models in our business. So first things first, let's talk about the tools that are available. Now, there are several, and we just saw today in the technology showcase, there's more and more AI tools becoming available. But what are some of the more staples that are being used, and especially as it pertains to writing with ai, which is what this panel is all about. Let's take a look at a list. So most of you in this room, probably all of you already understand ChatGPT. We have a tool called Opus and perplexity. We have Claud.

 Susan Theder (05:19):

Can we do a raise of hands for each? I'm curious.

Diana Cabrices (05:21):

Oh, what'd you say? Maybe

 Susan Theder (05:22):

Do a raise of hands for each one. See what?

Diana Cabrices (05:25):

I love that idea, Susan. Yes. Okay, so let's see if you use ChatGPT, raise your hand. Okay, awesome. Yep. How about Opus? Oh, we've got one guy over there, two over over here.

Mary Kate Gulick (05:36):

Going to love it. You're going to love it.

Diana Cabrices (05:37):

This one has saved me hours of work, perplexity. Nice. Okay, Claude,

 Susan Theder (05:45):

My second phase.

Diana Cabrices (05:46):

Yep, Gemini and then Dolly. Okay, this is exciting. We have some power users. We do. Okay, so we've got some good folks in this room here. So hopefully if you're already using these tools, what you walk away with is some inspiration on how to use them even better. Okay, now let's hop into content marketing. This entire panel is about writing with AI, and when you think about the writing aspect of your entire marketing game plan, you're creating content, but it often gets misunderstood, especially if you're not a content marketer. It gets understood. So I want to hear from my expert panelists here, starting with you, Susan, what exactly is content marketing?

 Susan Theder (06:28):

Well, it is something that I'm super passionate about. Advisors come to all of us probably and ask questions about what are the best practices for marketing? And most of you advisors would love to have an easy button, but the reality is, I think the thing that works best is content marketing, and it's a strategic approach to creating and sharing content that is a value to your target audience. So what's on their mind? What are their pain points? What are their concerns? Writing about that in a consistent way, and the objective is to attract, engage, build, trust, help guide them through a buying process and staying top of mind. All of that requires honestly, a lot of work because it requires consistency and content creation across multiple channels. But in a nutshell, that to me is what content marketing is.

Diana Cabrices (07:22):

Beautiful. Would you add anything, Mary Kate?

Mary Kate Gulick (07:24):

Yeah, I think I take something of a different view. I think it's a good definition of content marketing, but I've come to the conclusion that all marketing is content marketing.

(07:33):

At this point. I mean, you're starting with your ideal client persona, developing content around their questions, what they care about, using search data to determine what they're looking for so that they can find you and raise their hand and get into more traditional marketing, which still requires content creation that aligns with what first brought them to you. So the idea trying to divorce any of your content or any of your marketing efforts from content, I have yet to find a way to do that. So I started to think about it as all marketing is content marketing, but the core is to get them to find you and to engage with you in an inbound way. So I think that is so that you can move into a more traditional marketing model with them.

Diana Cabrices (08:19):

Love that. Now, Tim, how do you feel about it as a financial advisor?

Timothy Bickmore (08:22):

Our biggest pain point, Mary Kate and I were talking about it earlier, our firm at LBW, we're trying to figure out how to do content marketing, and I think as Susan said very well, it's a lot of work. And when you only have a small team and we're all very logical people and don't maybe have that creative side, we don't know how to create content or what content it is or what are these personas while we're trying to obviously service our clients, talk with them and have every other task that's part of our business. So content marketing is something that we recognize as a need, but also a lot of work in trying to figure out how to do it. And I do think AI can help, but as we probably continue to go through these slides, it also isn't just the silver bullet necessarily. It's not going to solve all your problems. And unfortunately, I think we've recognized that in our own firm thinking that it may have. So we're trying to get there, we're figuring it out as we go.

Diana Cabrices (09:12):

Absolutely. We're going to definitely dive into all of that. It's not perfect, and I'm excited to hear all of your perspective here because you use it all in different ways. So let's move along to our next question. Now, one of the things that people talk about the most when you're thinking about marketing and using AI is, well, what about originality? What's going to happen to originality? And as you use AI, what will happen to your message? What will happen to creativity? And so I think it's a genuine concern, and I also think there's an answer to it, and I want you all to answer the question, but how can advisors strike a balance between originality and scalability when using AI to create content in their business? Susan, I'll start with you.

 Susan Theder (09:54):

Oh, okay. I would say that it's a starting point. Anybody that's ever had somebody on staff that's written a draft for any kind of marketing material for them or hired a freelancer, you expect to make edits. I think it's the same thing with AI. I don't think anybody should look at it like, great, I'm going to put a prompt in and out comes my final product. It's a starting point, and the better your prompting is, the closer you get to the end point. But I don't think it removes your originality or your creativity because you're the one prompting it and you're the one steering it. And the better you do that, the more aligned it is exactly with your creativity and your ideas. But always, I think you all know, expect to take the time to personalize it and make sure that it is in your voice and all the data's right and all the rest. So it's never done when you hit enter.

Mary Kate Gulick (10:47):

Never done when you hit enter. But one of the things that I think really helps with this is start with originality. And this is from an operational perspective and not just about keeping things original. I'll start with a piece of content that's created without AI. That's kind of a pillar piece of content. So that could be a white paper, it could be a live presentation. I mean, it can even be a five minute video and we'll just use that as an example. Start with something that's all your words, all your ideas, that's 100% you. And then understand that as part of your content marketing strategy, you're going to turn that into different modalities, different formats to put it in front of your people where they are. There's no point going through the work of creating a great piece of content if you're not going to squeeze every last bit of goodness out of it.

(11:35):

So really make use of AI for that. Take the original piece and my go-to is I take that and then I use Opus to turn it into a million clips that I can use for social. I use chat just to put the transcript in there and give a very clear direction on how to create the blog post for me, how to create the landing page copy for me, how to create the emails for me. So it's using the source that I'm giving it, that's mine. And what it's creating out of that is new ways and new modalities. So that's really how I like to start. But I love Susan's idea of using it as a starting point. It's also great for ideas. Sometimes the blank page is your biggest enemy, and any start that AI gives you that you didn't have before is further along than you were before you started.

(12:23):

So if that's the thing that you need to get started, yeah, ask it for a 10 ideas for a great subject line. Ask for, Hey, this is my persona. What are five topics that they would be really interested in right now? We all have those days where we cannot do the creation just because we don't have the capacity for it. We're not in the head space for it. That's something that you can delegate just to give you that initial spark and sometimes more than that initial spark. And that's always a good thing and always a time saver.

Diana Cabrices (12:51):

Such a good take. I hope you're all taking notes right now because what is happening is this advice is sort of becoming a playbook for how you use AI to write in your business. So I see some of you taking notes. There's going to be a lot more ideas like that. So thank you Mary Kate for that perspective. Now, Tim, you have an interesting take on this. Being a financial advisor, how do you feel?

Timothy Bickmore (13:10):

So when ChatGPT first dropped back in 2023, I think everybody was like, oh my gosh, AI's taken over the advisor's role. And what I started to look at, I was like, wait a second, I am ChatGPT. A lot of clients ask me questions and then I give them answers. That's what AI does. And so when I started thinking about it, going back to Mary Kate's creation is sometimes it's hard to come up with what is the persona? Who am I supposed to be talking to? But my clients are constantly asking me questions, and if you're paying attention to it, you can start to find trends. And guess what we're already doing? We're sending emails, answering questions. So to give an example, the big hot topic is what do I do during the election? Should I sell out of my portfolio? Should I not?

(13:53):

Should I make trades? And you're usually going to put a thoughtful response for that answer back to the end client, but you can then take that and then you can put it into AI and then you can make it one to many modality. You can make it a TikTok, you can make an Instagram. And so I started to think about it in that realm and I was like, wait a second, we're creating content intraday multiple times a day, which is that originality. And then AI can help scale it, it can help put it out into it and go from one to one to one to many. And if you can start to track that data quality and say, okay, what are the common questions that are being asked throughout our client base, then you can start to understand what you actually want to give to the clients because they're asking for it. So we've really tried to dial in it and that range of things to make sure that we can get the originality and then the AI is helping do it at scale.

Diana Cabrices (14:43):

That's awesome. So what I'm hearing here is whenever you're in these meetings with clients or you're getting emails from clients, taking those questions, those frequently asked questions and using that as a guide or as prompts to add into ChatGPT for example, also from Mary Kate, love what you had to say about just repurposing, because often we have one really good piece of work, but it just sits there and after a while you think it goes stale, but it never goes stale. With AI, there's so many different things you can do. And then Susan using it as that starting point. That's something I do often in my business because sometimes as a business owner, your brain is just, it's off and you're tired and you're doing a million things. And that is why you have these tools to lean into them, to ask them for inspiration and then to sort of guide that kind of like what you were saying earlier, Susan, like edit and go edit and go, okay, so let's get real and let's talk about some pitfalls because there are pitfalls with anything. There's pros and cons to it all. So I want to hear from the three of you on what you feel there. What can go wrong with AI? Where can advisors who are starting to incorporate this into their practice, where can they go wrong with AI? Maybe I'll start with you, Tim.

Timothy Bickmore (15:52):

Yeah, I think some of the biggest pitfalls again is that everyone thinks it's going to be this answer to everything and it's not. It's really kind of the end point. Going back to my core and May's example, if you just drop the person without any directions or good data, they're not going to get anywhere. So you have to make sure that you're feeding it really good data. And honestly, with smaller RAs, like myself, data strategy is not even a thought in the process, but it is key to make sure that your AI works well for you. And then in addition, a lot of this is a big topic in general is compliance. What happens if I feed personal information into ChatGPT? Who's getting it, where is it going? Is that too much? Do we need to filter that out to make sure that it's not going into these LLMs?

(16:34):

Well, who's doing that? What's the workflow? What's the process? So there's a lot more to it to make sure that you can do it at scale without getting in trouble, without running into regulatory issues, and also to produce things that you actually want. So it's a great tool to enhance and to scale, but it is definitely not the end all be all. And I honestly say that from experience being like, oh yeah, this is what we can do. And then you start to get into the nuance of it and you're like, wait a second, we need to have more processes and controls around this to make sure that it's working for us and not against us moving forward.

Diana Cabrices (17:08):

Yeah, it kind of reminds me of the role of the CRM. Like everyone think it's the hub, everything goes in there, but if you have no workflows or no process to actually use it, what is it going to do for you? So Mary Kate, what would you add to that?

Mary Kate Gulick (17:22):

I think the big thing particularly in creating written content is that all large language models are subject to hallucinations, and it's because they're using human language and words are thoughts. And if it doesn't have all of the words that it needs to fill in the facts, it will create facts on its own. It's wild when you see it happen and you're like, where did this come from? But it does happen. And so the idea that it's, we can get lulled into this sense that it's just using information that I'm giving it, and it's just using information that's already out there. No, I mean it is, but it's also it's going to come across gaps and it's going to confabulate. So understanding that and knowing that it's not even just a question of having to edit, but before something even goes to compliance, there could be some non fact facts in there. So fact checking becomes more important than it is with a human user. And I think that's one of the big ones. And one of the things that more writerly types and people who are a little more traditional kind of content creators that they are very careful of.

Diana Cabrices (18:29):

Susan.

 Susan Theder (18:29):

I mean obviously all those completely true. And then in addition, when you're writing any kind of, let's say you're doing a blog and you want to reference data or research reports that have been done, I'll put in a prompt and ask it to write about, let's say trends in social security for the last 30 years or something like that. Sometimes it includes data and I've even then I'll say, can you add the source? And then if I'm in a rush, I might say, great, I'm done. Or I'll edit it and then I said, it's done. I'll send it to compliance. And they check the sources. And the sources are just lies. I mean just made up by AI. I'm like, ah, damn it. I asked and it answered what I really like about, I think I've got this right perplexity, and it gives you the answer.

(19:18):

It has a little bubble and includes the source with a link. So I think perplexity is the best for sources, but I keep asking it, you have to ask it, please write out the link. I can't copy and paste your bubble. So that's a step that you can save you some time. But another downfall chat has, you can attach documents, but with perplexity you can't. So each one kind of has its pluses. I've actually found myself, I'll do something in one of them. I honestly rotate around between chat bar perplexity and Claude, and I'll just sort of pick one. And then when I'm after three or four prompts back and forth, I might copy and paste it into, if it's something that is database, I usually go to Bard and I'll ask it, are there any points in here that aren't right? Or if it's more of a style thing, I might copy and paste it into Claude and say, could you write this any better? What would you do differently? So I almost use them against each other. I think they all have their own personalities and strengths and weaknesses.

Mary Kate Gulick (20:18):

That's very smart. And I think to the point about sources, you can give it sources that maybe there are only two studies that you want it to be taking from, and you can tell, Hey, you don't have to use the word, hey, but I do. I talked to it. It's my best friend. Please only use these sources and please focus on accuracy and only include data that's in here. And it will generally do that, but otherwise it makes assumptions and surmises.

Timothy Bickmore (20:45):

And that goes back to the prompting and the strict prompting is a very prompt. Engineering is a very, very real thing to make sure that you are trying to confine it and give it the right directions to give you the answer you're looking for.

 Susan Theder (20:56):

I do think prompt engineering is such a fancy word to basically just ask the right question.

(21:00):

.Yes,

Timothy Bickmore (21:01):

True. I love it though.

 Susan Theder (21:03):

Oh, not to belittle I just a college is now it's all prompted and obviously there is a more sophisticated, but as it relates to writing stuff, which is the session, I think, don't anybody be intimidated by needing to be a prompt engineer.

Timothy Bickmore (21:17):

That's a good point.

Diana Cabrices (21:18):

Yes. Great point. Okay, I love this. So pitfalls are real. You have to be careful. Check your sources, give it sources to use. I will tell you, I had to learn this the hard way. And finally after I was marketing some data and I realized, oh my god, this is not real data. So now if I do use ChatGPT, I'll go back and I'll say, Hey, can you please link those sources? And this happened maybe a couple months ago and it said, oh, we're sorry, we can't find the data where we source that from. And I was like, oh my goodness, I can never rely on these sources again. So some things we've learned the hard way. So I hope that this gives all of you some more just inspiration in how you're using Chachi pt. Okay, so let's talk about some best practices. Now, there's so many different things that we could cover here, and I think we'll keep this as succinct as possible because really we're going to show you this in action. But what are some of the best practices for writing with AI? Susan, I'd love to start with you.

 Susan Theder (22:16):

Alright, I'll be short because I'll demo it, but I think the first is establishing how to describe your voice so that when you do a prompt, if it's anything that you want in writing, whether it be a blog, email, social, whatever, it's writing to your voice, and I'll show you how to do that. Many of you probably already do, gosh, we could go on and on and on. I'll just leave it at that since we're trying to be short.

Mary Kate Gulick (22:37):

Okay, I'll be less concise because I'm not as considerate as Susan I have, there's a couple. My favorite is give it a role always. If I'm asking it to summarize data and give me recommendations, I'll say, act as a business analyst, read this and tell me what I can do with it that would be optimal to grow my business. Or I'll give it three of the things that I've already written and say, pretend you're the author of these posts in this same style. Give me 1500 words using this source. Always giving it a role really helps with your output. And then one of the other things is I love Susan's idea about voice printing. And I come back to this even going before that, the idea of your ideal client profile. So many firm owners and businesses, we think we know who our ideal client profile is or who we want it to be, and that's not necessarily who it is.

(23:33):

I challenge each of you download a CSV file of all of your client data, get rid of all the PII in it, take out the name, the email address, the address, the phone number, any account numbers, and if you have in their age account amounts, where they live from a city perspective, how many kids they have, whether they're married, birthdate, all of those kinds of things. And just upload that bad boy and say, please go through this data and give me a profile of who my average client is. You'll be incredibly surprised at the depth that you get and the rich picture that it can create of who your ideal client is and who your average client is. And sometimes you may say, this is actually true. This is not who I want my ideal client to be. And that gives you information about what you should be creating content about and what other things might need to change. So that is my favorite thing that I've ever learned.

Diana Cabrices (24:29):

And you bring up a good point just about your brand voice, your client persona, if you will. And Susan, when we plan this panel, you did have some really good points on that that I kind of just want to call out really quickly. If you don't really know where to start with that, right, how would you prompt a ChatGPT to help you understand that voice? Because people are always concerned this is going to produce something that isn't aligned with our brand and our brand voice.

 Susan Theder (24:55):

I mean, I'm going to show you up here. I would take a couple of samples of your writing, like any blogs that you've written, anything that you think does accurately portray your voice. It could be a bunch of copy from your website, maybe you even have brand guidelines, but you upload as many documents or examples as you want, and then you ask it to describe, and I'll show you using adjectives, your tone, voice style, even format, preferences and overall personality. And it'll come up with adjectives for each. I copied and pasted that into a document, and every time I do a prompt, I say, follow these. My voice print is, and I love your, that is a great idea about the client persona and figuring out all your clients understanding, I guess the breakdown of all of them too. You could take different segments and say, I mean it would show a lot, that's really interesting, but that's how you do it. And then you can save that, but you also, sometimes you're going to modify it based on the delivery of the message. Like yesterday, I was writing emails that we could give to our clients so that they could send out emails about Hurricane Milton. So I added and actually including giving it a role, I give it a goal. So my goal is to come across empathetic and comforting, and I'll elaborate maybe a little bit more right in full sentences, but I think your role goal and your voiceprint are really helpful.

Diana Cabrices (26:21):

Amazing. Now, Tim, as an advisor, not necessarily a professional marketer, I'm sure you market yourself well. What do you think about this? How do you feel about best practices and where to start and find your brand voice and all those things?

Timothy Bickmore (26:34):

It's great to hear you guys talk about the brand voice. I think that's what the scariest part is. I could go take some of our writing, but I have no confidence that it's the right voice. I have no confidence that it's the right thing. And as I think everybody knows, you put in ChatGPT garbage, garbage comes back out the other end. And so it's really difficult I think for smaller RIAs to know what that means and to have confidence that it's correct.

Mary Kate Gulick (26:55):

Yeah.

Timothy Bickmore (26:56):

And how do you find it? Because looking at two professionals that are like, Hey, we're experts in this. It's like, yeah, my website, I don't really like that much, so I don't want to put that in there. I don't think it's actually right. So trying to find that voice and understanding it I think is one of our key struggles is what,

 Susan Theder (27:12):

Does a great way you could do it is you could put the content in and ask it to describe the voice that you think it's portraying and say, what I want to convey is, and just spit out from the top of your head. That's the thing. Don't overanalyze what you're writing in the prompt and just say, this is what I want our brand voice to be, essentially. Tell me what adjectives I should use to evoke this in my writing.

Timothy Bickmore (27:35):

Which is great advice. That's probably something that I will go and do actually. So I appreciate that.

Diana Cabrices (27:41):

Real time. I love real time consulting. Okay, so we're going to get to one more question and then we're going to do the actual showing side of this next question is probably the last thing you want to talk about necessarily, but it continues to be a very important component of using AI to create content to put on the front lines of your business. So it is how can AI generated content remain compliant in such a regulated industry? I know FMG is coming up with a tool around this, so maybe we'll start with you, Susan.

 Susan Theder (28:13):

Okay. Well, an AI tool that picks up on compliance issues is basically what we're developing so that if you're on our platform and you're creating content, it's going to give you an alert. This could be, or that'll be a yellow or a red highlight. Definitely want to change this. And so it's just based on all of our learning, all of our content has to be compliant. It's basically just scanning it for, it's constantly learning based on our feedback as we make edits and refining itself. And then it'll prompt you when you're writing something that is potentially going to get flagged by compliance. And then we have the same tool for compliance to make it really easy and fast for them to review the content.

Mary Kate Gulick (28:55):

That's awesome.

 Susan Theder (28:56):

That's really cool.

Mary Kate Gulick (28:57):

And my whole thing is people, I feel like they get all up in their feelings about it, but anything created by AI is new content. Just like any other new content that gets created, it has to go through the same compliance process.

 Susan Theder (29:12):

I Know, right? I get why it's different.

Mary Kate Gulick (29:15):

It's an original piece of content that never existed before. It needs to go through compliance the same way.

Diana Cabrices (29:20):

Tim,

Timothy Bickmore (29:21):

I will tell you right now that our chief compliance officer would never not be allowed to review whatever we produce AI or not. So I think you're always going to have human in the loop, even with tools like FMG might be creating. It might streamline the process, help identify things a lot quicker, but I just don't know that a chief compliance officer is not going to review it or at least fact check it both for a lot of things that we've said here. There is hallucinations, it's real, is the content, right? Is the data correct? And I think from an advisory perspective, if you send that out and don't know it or haven't fact checked it, it can come back very quickly and not great. And then if that happens once, then for sure your chief compliance officer will make sure they're fact checking that every day of the week, at least mine will. I know that for a fact. So I think you'll always have some sort of human in the loop. I just think that AI will enhance the speed and scalability of what you can produce.

Diana Cabrices (30:10):

Absolutely. And you're right, I don't see it as very different. I think compliance is always going to be there to review the tool itself and once they can green light that anything that comes out of it, any of the content that's also getting reviewed. So the workflows are still there. Okay, thank you for all weighing in on that. Fun question about compliance. I say it's time to show. So we're going to head up to the podium here and we're going to show you a couple of different ways that we use these different tools. So we should, there we go. We're already live. So Susan, do you want to go up first?

 Susan Theder (30:43):

Me go first?

Diana Cabrices (30:44):

Yeah, sure.

 Susan Theder (30:45):

Alright,

Timothy Bickmore (30:46):

I'm excited. Susan,

 Susan Theder (30:47):

There's so many things I wanted done. I was making a list. I'm like, I think we have a little more time than we thought, but yeah,

Diana Cabrices (30:52):

See as you get that pulled up too, I'll just say like you just said, literally there's so many different things we could have done. When I was planning for this panel, we talked about client event ideas and getting that out of AI or webinar topics, subject lines, titles, social media captions, blog topics, image creation. So we're only going to cover just a little bit of that today. But yes, go ahead, Susan. We'd love to see what you have to show us.

 Susan Theder (31:15):

The typing's going to be a little off this keyboard. It's not my normal little Mac. Okay, you're fine. I'm going to show you how I might go about getting my voice print as I was describing. So I want to define my voice in terms of style, tone, personality, formality, sorry for the typing live. What is that style, tone, personality? Am I forgetting anything?

Timothy Bickmore (31:52):

Not the great person to ask.

 Susan Theder (31:53):

Style, tone, personality, general voice.

Diana Cabrices (32:02):

Yeah, there you go.

 Susan Theder (32:06):

To use in future prompts with you. So you get my style off the bat. All right, we're just going to go with typos because it doesn't care. That's right. Leverage this blog I wrote and use adjectives to describe each part of my voice. So then I go to upload, sorry for that long typing thing.

Diana Cabrices (32:44):

And yes, you can upload things directly there. So we have a Word document. It's on the desktop.

 Susan Theder (32:50):

Desktop. So I wrote a blog on navigating the costs of healthcare in retirement. So we're going to just, and there it goes.

Diana Cabrices (33:09):

Great. So let's break this down.

 Susan Theder (33:11):

So this to me, and I actually was practicing this earlier. I got the phone off is too much. It's more than I want. I just want something short that I can cut and paste. So I would say, I'm not going to take you through this whole thing, but I would prompt it to say, boil this into three adjectives for each of the key components that help you, that help direct you to deliver the most similar content to what I did in my blog. And I do that. If it was a little off, I'd give it another prompt, but that's how you get it. And then you just copy and paste that and I just have that in a document that's always open. So when I do prompts, I put it in. Honestly, I do feel like it learns you. I don't have to do it as often. So that's one. Then I'm going to say, okay, give me a question that if you were writing an email to your clients this week or next week, what would you want to write on? Anybody? Anything? Or do you want me to come up with it? Alright, I'll

Diana Cabrices (34:09):

Come maybe the hurricane.

 Susan Theder (34:11):

Okay, based on I did, I wrote a bunch of hurricane emails. I don't know, actually they were probably emailed out to everybody. One for if you are an advisor within the affected areas, and one if you just have potentially clients there, that was a great opportunity. And those timely ones are great for AI because they used to feel like, oh God, they're overwhelming. This point in time is not going to be timely tomorrow. I've got to get it out today. I don't have time, I missed it. So that helps. Based on this, write me an email that I can send to my clients. Now here's where we do the, I usually do goal, but I think I'm, I'm used to doing goal. So about, I'll just do on the blog I uploaded, I want to be informative. I also say a lot of times I say, and I don't want to say anything they already know.

Diana Cabrices (35:14):

Love that.

 Susan Theder (35:15):

I want to deliver ahas that make them say, oh wow, that's interesting. I want to tell someone about my amazing advisor. I am a financial, oh, the other thing is telling them who you are. If I'm writing on behalf of advisors, I always say I am a financial advisor serving. And then write your ideal client, high net worth individuals. Actually, let's not do it on the blog. Let's do it about the isn't this week, the cost of living increase for social security.

Diana Cabrices (36:05):

Sounds good.

 Susan Theder (36:09):

These live demos when you're typing are never quite as good as you think they will be.

Mary Kate Gulick (36:28):

Okay, let's see. I love that little subhead. What you may not know because you specifically asked for new information.

 Susan Theder (36:34):

So not bad does it achieve my objective. So I would potentially, I'd copy and paste this into chat, I mean into perplexity or one of the other ones and just say, is there anything that you would do to make this better? So just quick segues and other ways that I use it. A big one that I do is make this better. And sometimes that is literally my prompt, but my daughter might send me an essay and I just say, make this better. Like can you read it and proof it? And I'm like, oh my god, this better. And I'm always saying who I am, who it's for, what my goal is, is huge. So who I am, what my goal and your voice I think are the critical things, but make it better. Gosh, I use that all the time. And then version it for social.

(37:18):

I really am a good typer. But this thing is really weird. Using emojis and formatting that is best practices for social, let's say LinkedIn and start with a really good hook and end with hashtags and a question. And the versioning is huge. I mean, we already talked about it, but everything I do, I'm versioning for a blog into an email, into social, into you name it, into a podcast, into a video script. So I'd probably say so then almost. I don't think it's ever nailed it for social for me because that's just a totally different personality, but you get the gist. Other uses just, I'll throw out really fast. Oh gosh, nope. Taking too much time. Okay, upload any big research report you get. Or I just got a lease from my daughter's college apartment. I just upload and said, give me all the fine, tell me the things I need to know and anything that's got you. Or I'll upload a research report. I want to get stats on advisor's use of social media. It could be a 50 page research report. I'll just say, tell me all the stats about social media advisors use. It just saves so much time. So anyway, so sorry if I went long.

Diana Cabrices (38:39):

No, that's okay. That was awesome.

 Susan Theder (38:40):

That works, Susan.

Diana Cabrices (38:41):

That was so awesome. So Mary Kate's going to show us more of the visual side. So let's think about beyond just text videos for example.

 Susan Theder (38:48):

And I'm endless on this. It's

Mary Kate Gulick (38:50):

Oh, it's fun stuff. So one of the things we talked about was Opus Clip and I have just taken it and put a five minute YouTube video in there and I'm going to hit get clips in one click. And normally this would take four minutes, but I did it ahead of time. The goal of Opus is to basically, normally when I do a video, I would then make it into little clips for social, and that takes time. It's time that I don't have and it's something that I don't want to pay somebody to do. So Opus goes in and it develops the clips for you and optimizes them for social, and it gives you a score of, if you look over here, it gives you up to 10 different clips and it gives you a score based on virality. So if it has a good hook, good flow is providing good value and it's on trend, then it gives you some, and obviously you have the transcript, it cuts it up for you and it puts, I'm going to hit this one so you can see it puts the, it does all of your supers and things like that here.

(39:48):

And this is configured to my brand, so it has all of my colors and all of that good stuff. Very. So that's really fun. And all you have to do is download these and you can post 'em to social. And then I'll really quick just hop into Dolly, into Dolly Explore GPTs, right?

Diana Cabrices (40:07):

Yeah,

Mary Kate Gulick (40:08):

Thank you. And so I'm going to go ahead and give everybody a warning. When you're creating images, you have to let go of some control, and that's not an easy thing for everybody. So I just hop into the image generator and this is a good place to use prompt threading. So if I ask give, oh my goodness, you guys got to watch me have typos now me and it's, it's not your key, it's a different people. That's very good. They're sixties collaborating around an office table.

 Susan Theder (40:49):

Beautiful.

Mary Kate Gulick (40:50):

Great, easy peasy. And then it gives me a loading thing that's probably not going to be exactly what I want. And so from there I can start refining. So I'll say, okay, let's make this office more traditional dark wood luxury furnishings and make the clothing more professional and traditional. So maybe that very airy and modern space is not really aligned with who I am as an advisor. And so it takes what it's given me and all of the existing inputs about three people in their sixties, and this will be the last one we do because it is taking a little bit longer to load up here. There we go. That is a whole different kind of collaboration. Officer, you know what, that lady looks nothing like me. Make the woman have long red hair and a green shirt and it will bring up something different yet again, it won't take the same composition.

(42:13):

And so this is where I'm saying you have to let go of control. You can't be just change the wall color. And in open art you can actually upload photos and be like, use this person and put them in here. You can't do that with Dolly. Okay, there we go. That's a little more aligned with what I was looking for. So this is where we are going to start, but you can continue to thread your prompts to get closer to where you want. And then I'll say, this is great. Make this image into a social media graphic with the headline client collaboration perfected. And then it will take that and put an overlay of it. And it already knows what my types of choice are but,

 Susan Theder (43:03):

I find that it does tons of typos when you ask it to put text on a picture. Let's see.

Mary Kate Gulick (43:08):

Oh, it did it client collaboration perfected, but it gave me some weird new image. So that's something I would want to go back on. You're playing poker. It's like the dogs at the poker table. So not what I wanted. I would go back and be like, no, use this image and then you can just download that and use it for social. So that's a fun place to play, but again, you will never get exactly what you want. So we're not using it for things where you're looking for a precise match If you want to get into art AI, there's a lot more control in a tool like that.

Diana Cabrices (43:38):

Amazing. Thank you both so much for showing that. And then for Tim, I wanted you to showcase something that you're using in your business enough of the 60 year old's.

Timothy Bickmore (43:48):

So to give a little context behind why we have Tom Cruise giving a cocktail. So we as a firm, we have a really talented artist named Ying within our firm, and she was drawing birthday cards, hi, I'm drawn birthday cards for our clients. And it got to a point where our firm was growing and it was becoming a lot of work. And our clients loved it so much, and this is not a joke. We would go to their houses and they would have our birthday cards year after year being,

(44:13):

These are amazing.

 Susan Theder (44:15):

That's awesome.

Timothy Bickmore (44:15):

And so one of our core values was how do we keep that customization, that personal touch, but how do we do it at scale? And so what we decided to do once ChatGPT was released is, Hey, can we actually take those cards and can we digitize them? And so we actually now use AI to do our birthday cards for all of our clients on an annual basis. And what we do is in our meetings we'll find funny things that they talk about or whatever, and within our CRM, we'll say birthday idea and we'll put the idea down and then we create the image. So for example, one of our clients had said that they really wish that they could have a cocktail of Tom Cruise. And so we gave them a picture for their birthday saying, happy birthday, you're going to get a cocktail from Tom Cruise.

(44:57):

This client had just gotten a pet. So we did some pets and then we tried to describe obviously the dog that they just got, give them happy birthday. And I believe this was a travel plan for a client that they talked about. It was one of their dreams to go. And the best part about it is that now people can have it as an actual art piece. They can print it, they can do whatever they want with it, it's theirs. So it creates a little bit more, I guess, scalability and adjustability. So it's been a great transition. We were very nervous about it. Again, our clients loved our birthday cards, but they love this just as much or more, which has been a really nice.

 Susan Theder (45:31):

Can you describe the exact process you do to

Timothy Bickmore (45:35):

Create these?

 Susan Theder (45:35):

Yeah.

Timothy Bickmore (45:36):

So what we do, like I said from the beginning, we always are paying attention to our clients and client meetings. Anything that we catch in any meeting, we'll write down as like a birthday idea. So all of our advisors are doing that at all times, which is great. And then what we do is Ying on our team goes and takes those prompts and then she essentially has to be very descriptive about the prompt to create it. We use stable diffusion is the one that we ended up going

 Susan Theder (45:59):

Diffusion

Timothy Bickmore (46:00):

That we used. And being able to do that. And she has the prompts, I'm kind of the backup to her. So if she's out of the office or traveling or something, then I, for example, just did her birthday card for her. So that's essentially what we do. And it's very specific prompts, but the nice part is that because we writing down these ideas, however, doing people, the Tom Cruise one is very difficult. Sometimes you get some very odd images. So it depends on how descriptive it actually is. But usually we can dial it in to do it. And we also do style. So if it's artistic, not science fiction, whatever it may be, just depending on what we catch. Great

Diana Cabrices (46:39):

Idea. Wow. Thank you so much for sharing that. We were a bit ambitious and we thought maybe we'd have room for q and a, but we don't. However, some of us will be hanging around the conference. If you want to come up, ask us any questions. And the last thing before we hop off the stage is if you want to get connected with us, there's a few different QR codes here. For me personally, if you want to follow me along YouTube, I'm often posting content with marketing for advisors. Feel free to scan that. We've got Susan with FMG. Susan, does this lead to some content as well, like electoral, I think I saw.

 Susan Theder (47:10):

That. I think leads to a form and I'll reach out to you or just connect with me on LinkedIn.

Mary Kate Gulick (47:16):

Perfect. Mary Connect with me on LinkedIn as well that QR code leads to a research study on how consumers choose their financial advisors.

Timothy Bickmore (47:25):

Mine is just LinkedIn, so by all means, follow me. Happy to always chat.

Diana Cabrices (47:29):

Thank you so much to our amazing panelists. Everyone in the audience, thank you as well.