Transcription
Hey everybody, this is Matt Howerin from Zox. I'm the chief evangelist and we have something for you that is really quite groundbreaking. And I'm not saying that is a kind of a tongue-in-cheek sort of thing.
Uh, we had this first webinar that uh we went over some MCP stuff and uh we had overwhelming response to people just really wanting to know how to do it. So we took it back to the team and the team well, they started building stuff so that we could have Mark Gilbert who's our CEO and we've got Neil on who's an actual adviser just like all of you who has built an unbelievable MCP protocol that he's going to share a little bit later.
But a couple of very quick housekeeping things. We are recording this. So if you have to drop, just go ahead and drop. We're going to send you a recording afterwards. And also, since we're sending you the recording, if you know another financial services professional that could really use this, please make sure that you share the link that we're going to send with you, the recording, to them because this is really a practice-changing thing.
We're going to run about 45-50 minutes is the ultimate goal here. And if you have any questions, please put them either in the chat or Q&A. Uh, we've got a couple of people on our team in the background who are going to be managing that chat and they're going to bubble them up to um us if we have any big picture questions.
Um, so Mark, I'm going to just turn this over to you, my friend.
>> Awesome. I appreciate it, Matt, and thanks everyone for your time today. Um, we're really excited. Um, some of you may we may have already chatted with uh or heard us talk a little bit about MCP, but we're going to start from the start, go through this fairly quickly, but happy to dive in, you know, wherever uh I guess in a few areas. And I think more importantly, more exciting than me and even more exciting than Matt, we have Neil uh who has been using this um you know, in in his I would say day-to-day practice. And that is uh a lot more interesting. So we're super excited to show you that.
Before we hop in though, you know, one of the questions we get all the time is really kind of like, what is MCP? What does it do for me? Why should I use this thing? Uh, MCP is is actually a very technical acronym in the sense of like it's essentially APIs between um AI systems is probably the easiest way to think about it. So just like in the past when we said, hey, system A integrates with system B, um, that would be using something typically called REST APIs on the internet. There's a few other ways of doing it, but that was most common.
When we look at AI systems, um, we often will use something called MCP, which is just an easy way for the AI systems to to communicate. What it really means to you is that we actually with Zox now can get all of the rich context that is inside of Zox about your clients um, about any of these things and get that onto uh, into the other AI systems that you want. So today um I'll walk you through using this with Claude, just an Anthropic product that a lot of people use. Um, we are also in the uh OpenAI, the ChatGPT library. So if you use ChatGPT, everything I'm showing you here looks very similar in that world and you can go use it too, whatever you want to use.
And for those people using Copilot, um, I believe we're going to put this in the chat a little later or in the questions and answers, but we have the ability to um uh actually install it through Copilot Studio. So essentially an admin on your I'll say on your Microsoft admin accounts needs to go in and install it. It's frankly not the most straightforward thing, but you can do it a little bit. And then Copilot also has has access to it as well, which is great.
Um, and I'm seeing a question already from Gary and we'll cover that a little bit just around kind of like privacy and that's a huge piece to hit um, as well.
Um, so at a high level with MCP again, what we can do and and the examples we'll show you today are really hooking it up to Claude. So what what's happening there is instead of taking, you know, like transcripts or all your notes from stuff and copying and pasting them around into these AI chatbots and these AI systems, you actually can just access Zox data directly in that and do a bunch of really cool things. These systems also allow us to put in essentially like, you know, templates and projects and all of these things that make this a little bit easier for you. So, we're super excited to show you um what that looks like. And you can think of this really as a bridge to connect the data from Zox into the AI system in a way that is governed, that is secure with that connection, that respects the different privacy pieces. And what you'll see more and more is that we are going to give you more and more controls essentially on um, you know, more and more controls essentially on on what that uh what data is going into the AI systems and what you're allowing it to see in that. So super excited about that.
Um, and with that, I think we'd love to get to the next slide and hop into a bit of a demo.
Awesome. So, what I'm going to do is share my screen. I think I'm up for the first demo and then Neil's getting the lion's share of this. So, I will, as I mentioned, keep mine fairly quick so you can get to the more exciting stuff.
Um, but I'm actually going to show you two quick things. Um, for those using Zox already, this is Zox. If you go down to the integrations tab and then click on Claude, uh, sorry, on integrations and then click on apps. So when you get to the integrations tab, you see everything here on the apps button. Um, you can go in and connect, manage. You can also go in from Claude and you can see from ChatGPT, you can do it directly from their library to connect in, but this is essentially where you manage things and you can look at what access it has, what permissions it has, and all of those capabilities as well. So it's quite rich here.
Um, but more importantly and more exciting, I just want to show you what this looks like once we connect them. So, this is in Claude. I actually have a a team demo account with Jim Hardman where we do a lot of demos together. So, you'll see his name pop up in my account because we share things. And just like with Zox in uh in Claude, you can actually share across your teammates and do those things as well with projects.
And so, what I'm going to go in and do is I'll actually just start by asking it um I want to make sure it's hooked up. Can you see Zox data? Load. And it's going to go in and ask uh and look and introspect and I just want to make sure that the connection's hooked up. And this is a way to tell that MCP is working for you.
Um, and so it's going through and essentially like interrogating what it can do and what data it has here and says, yes, I can see your data. Here are the different things I can see. Here's some meetings that you have and everything else. And as I mentioned, this is uh Jim and my and I share this environment.
Um, and so then I can go and click in here and I can type in anything I want and ask it. So I can ask it to fill out forms. I can ask it for, you know, any information on things. There's some really cool stuff here. One of the things though that we also do is we have this concept of connectors and skills. So Zox is connected here. Just as an example, I also have Box connected, which is a file share that a lot of people use. There's a ton of connectors uh both for Claude and ChatGPT. So I would recommend you are welcome to like explore those and you can use them together, which I'll show you a quick example of in a second too.
But then I can go in and say, hey, I want to add this from Zox and you can see we have a number of I will say kind of default examples here that are really quite powerful. So I'm going to pick an annual client letter. It's going to ask me some questions like, who is this for? So I'm going to say, hey, I want this letter for John and Michelle Smith. Um, I prefer structured HTML as an output instead of PDFs and I love warm and personal. And then I'm actually going to add the prompt.
And what this is going to do is it'll actually build the prompt for me. And you can see here it gives a whole bunch of context in in um in a little file to add. What I'm also going to do, and I'm going to cheat a little bit here, just like a baking show, is I'm actually going to paste in a bunch of bullet points that I've used before um with this that I've tried. So, I want to create a personalized letter. Should use Zox intelligence? Provide an overview. And I'm going to hit go.
And it's going to start thinking and asking uh and telling me frankly what it's doing. But the interesting thing here is that it has accessed to uh John and Michelle Smith in this case. And it's actually starting to build a letter uh for the whole year, like an annual client letter. So, it's essentially looking at every time I've asked, I've met with them, you know, what are their goals at the start of the year? What happened at the end of the year? What were the things that we accomplished with them, which is awesome. What are the things maybe that we're focusing on next year? You can see it's going to work through step by step. It's starting to get through contacts. This is going to take a couple minutes and I won't make you sit through it.
So, I ran it earlier today and I'll show you uh what the results look like and give you a few other examples, but it's just really interesting. Here you can see if you go through all of this, all of the different things it's doing. If there's multiple contacts, it'll ask you which one you want to use it for. All of these things as well, or if it needs other information to solve out.
Um, so I'm going to go in here and I'm actually just going to open up a previous one where you can see I did the same thing. And as I scroll down, it'll tell me all the all the different data it put in. This took about four or five minutes to run. And then because I asked for an HTML, I'm actually going to open it up in Chrome.
Um, I don't know why that's popping up on my other screen. Apologies. But what happens here is it'll basically come in and says, "Hey, uh, Jim set our profile template as Jim Hardman Financials." So, you can also go in and say, "Hey, this is what our brand looks like for our firm. This is, you know, our our letterhead, all of these types of things." And it'll it'll match to it. And then it basically goes through everything from where we began, the year that changed, you know, what happened, all the different events, and actually put a chart in around like how that plan is coming to life. It's quite powerful.
And, you know, from a lot of the advisors we talk with, they would love to send these to their clients every every year. It's a very powerful thing to stay connected and frankly remind them of all the progress they made on the journey, which I think a lot of you, you know, know feels slow at times but matters as it compounds and um and really kind of walks through all this. And then the nice thing is you can iterate in in the systems too. So if you're in ChatGPT or Claude or Copilot, you can say, "Hey, make this more formal. Make this shorter. Make this longer." Whatever you want to do, and it'll actually regenerate it for you.
Once you get it down and you have something you like, you can then reuse it for your other clients, too. So say, "Hey, don't just do it for John and Michelle, do it for Matt, do it for Neil, do and it'll just basically go in and start to produce all of these for you." So it's super powerful.
Um, and frankly allows you to uh have a very personalized touch with a lot of people um in a way that is uh just, you know, frankly doesn't take you 10, 20, 30, 40 hours to do, which it would normally take if you were trying to do this across all of your clients.
Some other really powerful examples are you can actually ask it as well for even coaching. We have scorecards in Zox around conversations, but you can say, at what point in this conversation did they seem to engage or not, or where did they go? And it'll actually go in and analyze that for you and even show you like, you know, when did they start to share very personal information, which is a big sign for like a client, you know, or prospect starting to build trust in you.
Um, another quick example of just how these two systems can work together is I asked this to use I actually just dragged in a PDF, which was a Schwab application, account application. You can fill out forms, you know, if they're in Zox and Zox. There's, you know, as everyone knows, a ton of applications in the space for almost endless account types. And so, you can actually drop PDFs in here, too. And what I did is I just said, use Zox data for Peter and Paul Reynolds and uh, combined that with the information that they shared with me in in the Box account.
Um, so I was using Box here to like, you know, where they they would upload information. It actually went through both, filled out both, um, pulled in all the information. and it will tell you if anything's missing and then filled out the PDF to that and I can just open that as well. So, it's very powerful.
Again, just at a high level here, the big things that are happening is because all of the context that you've had with your clients, not just from the conversations, but other attached systems, we, you know, we we understand more and more about them. We've built out a profile. One of the interesting I'll just say kind of like tech uh tech things here that's quite powerful is when we actually uh see the MCP uh interface get used, over 90% of those queries that were asking it actually never go down to transcripts. So, if you are um, let's say at maybe a a firm that that doesn't hold transcripts for very long, like maybe a week or so, which quite a lot of the frankly the broker dealers do and others do as well, a lot of this is still I would actually say the vast majority of this is still available to you because the vast majority of questions and interactions we're seeing are going to this client profile. It doesn't actually have to go down to the specifics of like words uh from one conversation or another.
So, that just gives you a quick overview of kind of how to get going, you know, how it shows up in Zox, how you can see the permissions. We did get a question on kind of the different versions. And maybe I'll save that a little later, Matt, if that's okay with you, and we can hit the end of like what are the versions both with OpenAI and Anthropic, so Claude and ChatGPT that people are using from a data security settings and what do you want to do there?
Um, and we'll talk through that. But, uh, I will say on that note, I'd pass it back to you, Matt, and, uh, we can get into the even more interesting stuff.
>> Yeah. So, we're going to jump into the fireside chat now. Uh, but, uh, I I got just a couple of quick things. Uh, number one, um, you So, so Mark exported that as an HTML. You can export it as a Word document or whatever, too. So, if you want to be a little bit more hands-on with it instead of iterating the prompt, uh, you can do. I just did that on my MCP with with my my Claude. So that that was absolutely fantastic.
Um, and the other thing is I'm devastated that you just told me something about a baking show I didn't know, Mark. And so now I don't know if I can ever watch those baking shows again. Uh, so, yeah, you just this rough Wednesday now for me. All right, Neil. Well, so you're an advisor using this in your practice really on a regular basis.
>> When did this epiphany hit you? When did you realize that you could use this and it could be practice-changing?
>> Zox in general or the integration with with AI?
>> Well, I mean, you could do a shameless plug for Zox. I appreciate that, but let's talk about the MCP.
>> So, I I sat through the one of the first webinars you guys did with the MCP side of things and saw the letter and whatnot and really just kind of started playing around with it. And I've started using Claude Co-work, which is on the paid version that resides more on your actual workstation than the cloud version does. And the ability to that it has to look at other files and incorporate that into the meeting prep thing that I've kind of developed uh on my own with that. So, it was kind of one of those things where, wow, the letter output's great, but what if we do this?
>> And I just kind of went another step with that.
>> Well, and let's talk about what if we do this, Mark. You when you guys made the decision to spend uh, you know, a lot of our our programming team's time on on building this out, what gap did you see or did you hear from our advisory board or the advisors you've talked to that you really wanted to solve that gap?
>> Yeah. So, we see a few different things and I think Neil will show some of this today. Like, you know, there's a lot obviously as we continue to build Zox where we see it used quite regularly and we want to make that very easy from Zox, right? So, again, we can fill out forms, pull all this information, all that. Um, even with all of that, right, some people want to do things either that we can't do yet or frankly may never put in the system. So, like the example we just showed of um, you know, like a tailored annual letter, something frankly we've wanted to do for a while, we just haven't had the time. I think Neil's going to walk through some just interesting things where he really cares deeply about the exact tailoring, right, of how things look and feel for the client, which makes sense. And it just gives you like full control over that. So, I would say that's one area that's pretty big.
We see another area people are using it more to do analysis. So, as I mentioned, like they may say like, hey, like why did this client go well? Why did it not? More at the firm level if they want to start asking questions, you can, you know, like these these broad LLMs and and AI systems are extremely good at analyzing a lot of data and so they can start to find patterns for you and everything else. And then the third one, which people aren't seeing yet, although we actually had a great announcement last week uh I would say leading towards this quickly, is, you know, as you're seeing more and more AI systems come online um and when I say that like, you know, existing software use that are adding AI capabilities or new, excuse me, new types of software with it. Those also will expose out through MCP. So, we announced, I believe it was last week, uh, Conquest, you know, as a as our, you know, uh, very recent partner. Um, on the planning side, it's used heavily in Canada. We have a lot of Canadian customers now, which we love. And, uh, we're seeing it used, you know, as well in the States. It has a lot of AI capabilities built in directly. And so, you'll start to see more and more, um, come out there where you can ask things in Zox that will go directly into Conquest and vice versa as well. And so one of the nice things is that people can really sit in whatever UI they want. I think people are used to like I go to this app to do this, I go to this app to do this. And that makes sense in a lot of this now. You can, you know, you can do some stuff in Zox, but for you because we're really, >> frankly, breaking down these data silos that have been such a a challenge, frankly, in the industry.
So Neil, I think a lot of people uh expect the world uh from AI now, which is a little bit maddening for those of us who've been in financial services for a long time, that this is just one of those things that they think it's like this magical panacea, this wonderful uh fix all of everything. It and we we know that it's amazing, but it doesn't fix everything. When you were working with this, what really surprised you? like you're like, I don't really know if it's going to be able to do this, but ding ding ding ding ding, it did do these things. So, can you give us a couple examples?
>> Yeah. So, the the meeting prep skill that I developed was more something that Zox wasn't able to do on its own from pulling information from the other two systems I use, which are Redtail and Orion. And we have a very specific agenda and template that we kind of use for our our annual review meetings when we sit down with our clients. And it's really useful for me at the end to have a memo that's has all these things on it. So I can just quickly go and, okay, 10 minutes, I'm ready to go for the meeting and I don't have to spend a bunch of time looking things up. And Zox was not quite able to pull that exact information from Orion. So when I started messing around with this and and some other experimenting that I had done outside of Zox with the AI and having it be able to read from a spreadsheet or read from a database kind of made me think, oh, this is this maybe will work. So it was kind of a combination of those two things that I kind of started down this path and that's what led me to where I am today in this regard. So it was it was kind of just filling a need that in experimentation that got me here.
>> Well, and I really do highly recommend that you experiment with non-private client data, right? Just go go in and mess around with this. I was just playing with uh with my my Claude with the the different integrations. Um, and I was really surprised at what it was able to do, what it was able to integrate with the the programs that I have on my computer.
Um, just a a fun uh side tangent here is I've built a a daily briefing, which is a pretty standard thing where it will pull from a bunch of different sources and every morning all I have to do is to say to Claude, "Good morning," and it will populate this very, very nice thing that talks about what appointments I have, email that I have to, you know, Slack messages, Asana tasks, those sorts of things. And it also pulls some relevant news, like, you know, who won in the basketball game last night. You know, it's a lot of fun to just be able to get that because, you know, that's kind of what some of us really wanted from our robots that aren't here yet. Uh, so, you know, the robot just walks in and says, "Good morning, Matt. Here's your day." And so, now I have Claude doing that.
But >> let's talk about your day because, you know, I do my daily briefing every day. What is a day in the life of you using um this this system that you've built look like for you? H how does it change your workflow? What has it done for your life?
So basically this workflow has cut down meeting prep for me and my staff. It's not just me, but me and my staff probably from 30 to 45 minutes to about 10 minutes to to do the workflow to get ready for a meeting. So I can, you know, if I'm super busy and have appointments scheduled, you know, stacked on top of each other, I basically have this one to two-page memo that I can sit and leaf through, find the things that I want to talk about and can step into that meeting without having to spend an hour or, you know, 30 minutes to an hour going back through old notes and looking for certain specific tasks. Talking points. You and and the non-financial talking points is a lot of is what we use a lot too just for conversation starters, right? So, it it helps consolidate that all into one place and I can sit there with that on my desk during the meeting and and just refer back to it and use that to to get things started and it's it's cut down on some time that it takes otherwise to go through that process. Being being in the industry for 20 years and and working with offices just like yours, Neil. I mean, that meeting prep part of this uh always used to be super stressful for for our advisors. They were terrified that they were going to walk in and miss something and and say something that wasn't 100% correct or forget a birthday. I mean, there's just simple things like that that now you can go ahead and pull. And I just think that's amazing.
Okay, but now now here's the best part. You're going to show us something like you're going to give everybody on the webinar or something like sink their teeth into. So, take it away.
>> So, uh, when I look at uh when I look at my meetings for that day and I kind of just jump into things, I come to Claude and I have a bunch of different things set up here, but under my projects, I have a project that is just meeting prep. And I simply come in and type meeting slash meeting prep.
>> Oh, can you share your screen?
>> Oh, I I thought I did share my screen. Ah, my mistake.
>> And that's why I am here.
>> I got part way, just didn't get all the way. So, I go to my Claude project in this and then I just come in here and type meeting prep memo and typing in real time is always fun. And then I type in my client's name and hit return. And off it goes. And it's off and running. And it's going through. And it will I have not been so comfortable where I just do always allow on the permissions it asks for various things. So it will prompt me to give it permission to access certain directories and certain things in the Zox system. And it goes through and reads some of the and we'll talk about these markdown files here in a minute. And I saw coming up on the uh on the chat, what is HTML? Well, get ready for something even more new, which is the MD side of things if you're into the super nerdy side of this. But basically, then it goes through and it runs this skill that we've built. And by saying we, I mean that Claude and I together collaborated in kind of an interview process. And Claude really built the skill based on what I wanted it to put out there. I didn't write code. I didn't go in and design anything. I said, "This is what I want." And Claude figured out how to do it, which is really wild. It It took me about an hour or so to get it built, but it was time well spent to invest in that. And that that's how you have to view it as an investment as as such.
So, as it's run, it kind of runs through all this stuff and and goes through that. But what it generates at the end of the day is this nice memo that comes up. And I've redacted this one. Uh, and it gives me a summary of all the things that I really want to see during the course of a meeting in in advance. So, one of the things our fee wants to know is, do we have it assigned to the proper model for portfolio management? What is the fee schedule that we are charging on that client? We have to verify that that's correct. So, this is able to pull from an Orion spreadsheet that I have set up and stored on my local hard drive and it's pulling this information. I also want to know what's the performance done year-to-date. So, if there's any big surprises, I can at least look into it or dig into some of the details as to why things are doing well or doing poorly. And then it also talks about required minimum distributions and we use the RMD side as a significant planning opportunity for qualified charitable distributions primarily, but also to understand the tax consequences of that as well.
And then we get into other things where we have information and this is pulling from the Zox information in particular the family information. This particular individual's son was a senior in high school. Last time we met, he was considering a bunch of different schools. Uh, personal interest and hobbies, personal characteristics and the communication style, other summaries, key discussion points that we've talked about in the past. And probably my favorite top my favorite section is the conversation starters. So, we can ask about his son's college choice and what's going on uh and and other things that we've talked about that we had dealt with in the past that maybe weren't completed either through us or through the client not providing or following through on the things that they said they were going to do because as much as we try to do stuff, clients are always great about getting their homework done too, right? So, we have this great memo and this one, this particular client's a very detailed situation, and it's more pages than normal. It ended up being about seven pages. And I think they're going to make this available to share so someone can see as an example. And again, I've kind of taken all the personal information out of that so not to jeopardize my client in this situation.
But it basically starts with when we're doing this with these the the key thing here is these three spreadsheets that we have. And that's the Orion account data, the performance by registration data from Orion, and the RMD tables from Orion. And basically once a month, I go in and download the Orion account data and the performance by registration. I redact certain very important things. I delete columns out of that spreadsheet, which are account numbers, social security numbers, addresses, dates of birth, anything that would be considered PII. I delete those columns from those spreadsheets. Takes a couple minutes but it's there for the rest of the time. Things like the Orion household number only means something to Orion. It doesn't really mean anything to anybody else. Nobody can really use that.
Uh, and then it is reading that file, those files, Claude, and then compiling that information along with the meeting notes. And this is the thing, this is the gap that it filled in the Zox system, right? Zox wasn't able to see that Orion data. So now it's able to pull that in. So I have something that I'm ready to look at right then and there.
Okay. So we are we are going to share that. I do have to say that how long this document was was a little bit eye-opening for me. Um, but but let's let's I I want to take this a little bit further and I'm going to warn everybody. We didn't really go over this in in the prep. So Neil, I'm just going to ask you on the fly here a little bit, but when it comes to um, so how early are you doing this in the process? Uh, so like, is this five minutes before the meeting? Is this the day before the meeting? Can you talk a little bit more about the practical usage of that document?
>> Yeah. So usually we'll run that first thing in the morning when I look at my calendar and see I've got these appointments today. I'll run that for each of those appointments and then I, you know, I'll print them out and that again, let me I'll go back to that document being that as long as it was, most of them are about two to three pages.
>> Most of them aren't seven pages. That that one is just a it's a extreme example I guess for lack of a better term, right? It's not it's not the standard output that I get. It most of them are two to three pages. I'll run these in the mornings when I come in knowing that the I have these three meetings today, need to be ready for all three of those meetings. I'll have them and I'll have them printed out when my focus can turn to that meeting, I can grab that memo and read it, go through it, take it into the meeting with me. I don't hand it to the client. It's internal use only kind of thing. And then I can scribble some other notes on it from what I might remember. Not everything's 100% correct. So I can kind of filter that out as well. Uh, in in that particular memo, it says the client's an Ohio State fan. That's not true. I'm the Ohio State fan, but Zach somehow has got that confused.
>> We'll get we'll forgive it, right? But you know, and so we'll do it in the mornings before, you know, to start the day is kind of how I look at that.
>> So if you're an advisor who who but you can run this 48 hours before so that either your team could go at and that that's my next level question. Um, I I did a bad job with this and I'm sorry. Like everybody knows who Mark is. C Neil, can you just tell everybody just a little bit about your practice makeup and like do you have a niche or somebody you serve specifically?
>> So, we don't really have a niche. We we are in a small town in southeastern Ohio. Uh, my team in particular is a we I have two advisors that are kind of learning underneath of me. One of them has been in the business for all of five months. The other one's been around for about four years now. Uh, and I have an office manager and trying to get to the point where I feel comfortable enough to do this at an enterprise level where I can turn this over to my junior advisor and say, "Hey, do these meeting preps through Claude." But that one that involves an investment in the in the Claude side of things because I have to have the enterprise version, which isn't it's not ridiculous. It's $1,200 a year. Not awful. And, you know, to roll it out to the team and then have figure out the the internal structures of how to store all that information uh behind the scenes, that's using the Co-work. And and I want to stress that again, that it's it's not I'm not throwing this out on the web. It's all done inside that Claude Co-work thing. And I'm not a good enough expert to explain how Co-work is different from the website things.
>> Well, but but you you said something earlier too, which I also want to highlight, which is this was an iterative process and and I can't I love that you allowed Claude to help you build the prompts that was needed in order to make Claude work more efficiently. Did was that something that you had heard that you could do or were you just messing around with it like, oh, it's telling me to do this?
So, I've used I've done that with other things, okay, in Claude, just messing around. And, you know, I'm that kind of nerd, right, that will get into this stuff. And so, basically just kind of said, here's what I'm looking for. Let's do it. Here's where the files are. Let's do it. Zox, I've got the connector set up for Zox. Let's use that, >> you know, and then ask me whatever questions to clarify the output. If there's something that is not clear, let's do this. And then let's build the skill. Let's build the context files. Let's build the uh main things that you're going to refer to every time we do this.
>> Okay. Hey Mark, uh, the the chat really blew up with something there. I don't know if you saw that. Uh, would you mind talking a little bit about the Orion integration? Uh, because it seemed that that was a pretty hot topic.
>> Yeah, there's Sorry, I'm just trying to scan the chat and see if I got all the context on it. So generally the Orion integration will pull quite a lot of data um from Orion like like quite broadly. Um, a simple way to look at it is if you can go ask inside of Zox like if you can go ask the AI assistant if it can, you know, like for let's say a data point, let's say Matt, you're my client, like tell me this thing about Matt um that's there. By default, for data security reasons, we don't pass all of the data in the connections all the way through, frankly. So I saw another question, apologies, we haven't had a chance to answer yet around Redtail of like, can I access every data point in Redtail? The way to think about it is like we essentially look across all of these pieces and see it. So if you're keeping the data in Zox around the client and everything we kind of know about them and use them, we'll we'll have context on that, but you can't, you know, necessarily go down to each point in each of the connected systems and grab them. It's not like when we ask, we'll go to everything and go. We essentially are looking across different pieces. There's a few different reasons for that from data security to also just frankly the data rights of u of the different apps that we use. And so that is one of the, you know, to the earlier questions Matt that you asked around like, you know, when when do we use MCP, why is it, you know, through let's say through Claude uh versus in Zox directly and things like that. There are things for a number of reasons that we just won't um see. Like as an example, in some of the systems, there's very rich, you know, PDF reports that they produce that you can't really like pull into Zox yet. Um, you can take those and frankly drag those in. Again, be cautious and we'll talk more about the data security pieces. There's definitely a number of questions in there around that.
>> Uh, but it, you know, like like I would say, you know, like the the broader LLMs are very good at analyzing those. So, it is >> a bit of a uh a way to kind of like work where if the integration isn't directly or you can't get directly to that point, it'll just basically go through like you can always drag a PDF in and it does quite quite well at understanding that.
Okay. Well, I mean, and not not to just jump ahead because I know the team hasn't worked with me long, but I'd love to do that because can we just talk about that real quick, please? I mean, I really want to talk a little bit more about because I know uh the one of the things and Neil, I I think you've heard this, Mark, you hear this. I hear this at conferences all the time is people are really concerned about the security components of the data. So, could you elaborate on that please, Mark?
>> Yeah. So like the way I'll just I guess back up a little bit and then maybe we can talk through some of the details on it now and what you want to do there uh or or just, you know, what the options are for everyone. But essentially at a high level um, you know, when you connect two systems together, that second con that s that second system sorry can see a lot of data from the first system just straight out. And I say that in the sense of like, you know, just like, you know, we have a lot of people let's say connect Zox to e-money. They, you know, hopefully trust Zox. We've been reviewed and approved. They trust e-money. They've been reviewed and approved and that makes it feel a lot, you know, like frankly more comfortable for good reason um when you connect in I'll say Anthropic or or OpenAI, like ChatGPT or Claude um into Zox, it can see a reasonable amount of information, right? Anything I'll just go through it because I I think people kind of need need the details here frankly, you want to be able to explain it to your clients too uh if you have us redact anything coming into Zox, that that data is still redacted and will never get pushed through, right? So, let's say if someone is set up to redact social security numbers, that redacted, you know, will go the whole way through and everything else. Um, if people delete data from Zox, that data is not going through. So, there's like one question around like, hey, if I delete meetings after I push them to Redtail, like, do you still remember that? And we don't, right? Like when you hit delete, that data is gone. So it it basically pushes the whole way through.
Um, when we do query, we can query into Redtail. So there may be some context, like let's say from that meeting note that's coming back, but we as Zox don't hold it. When you're looking at and I would say the three most common ones we see frankly are Claude, uh, ChatGPT, and Copilot. When you're looking at those and I'll start like Claude and ChatGPT are uh pro probably bigger ones. The things to look at there are are they training with your data and then also are they storing your data and if they're storing your data for how long. And so you can look at the different plans. Um, different areas frankly have different plans for Claude and the way it works. Claude generally does not uh train with any data that comes from you. But you can go in and check whether you want an enterprise version or a business version. In any of these systems, typically once you get to the paid version, you can tell it, you know, it it won't train by default uh unless you opt in. I know with ChatGPT that's also how it works. You essentially like if you're on the free version, assume they're training with your data just straight out. They may not or they they say they may have the right, but I would frankly get to a paid version because that is is always better. You just have more controls. In addition, you can tell it not to store your data. Generally, these systems will store your data for 30 days more for frankly like kind of their own compliance and checking.
Um, one important piece for those, this is really in the weeds, apologies, but um, uh, Mythos or what what is now, um, I think it's called, uh, Fable is the, uh, Anthropic just released this model that was kind of the, you know, big one. Uh, sorry, Fable came out yesterday. And so for people using that, um, with Fable, there is no zero data retention option. So frankly for security reasons and a variety of other reasons, they are keeping track of what you do for 30 days. Mostly I'll just, you know, my words not theirs, but mostly it's so people aren't misusing the model to do cybersecurity things and bio things that we don't want to do. That's why they've done it, but you just need to be aware of that.
>> And then um with Claude and ChatGPT and like OpenAI um you can request on the business and enterprise accounts to have what they'll call ZDR, which is zero data retention. And what that means is that there is no storage of any data. They essentially pass it through and go out and it goes the whole way through.
Um, in the systems like in Claude, if you go in and look at if you see your history, it's remembering it, right?
So you can go in and delete it. You can go and tell it not to hold history. Um, you can also, I know with Claude, I don't know with ChatGPT, tell it not to process PII if you want. So if it does see it, it essentially won't store anything in there, and they have that ability to.
Okay. So I appreciate that. Thank you for getting into the weeds, cuz it seemed like people really wanted us to dive in there a little bit. And, um, one of the really great tools that we have, uh, is, is you can actually call and talk to humans. Zox can answer a lot of these questions for you. So, so please make sure that you do that. Um, you can actually go into Zox and in the lower right hand corner, uh, you can go ahead and pop that up and ask a question. And, um, and not only will you get an answer, but if, if need be, we can go ahead and have our service team, uh, respond to it.
Now, Neil, you said that that took you an hour-ish to go ahead and build all of that, which, by the way, that seems flipping insane to me that it only took you an hour. But if you, I mean, you're talking to, you know, 500 people here. Um, I'm assuming that not all 500 are using this. If you were going to kind of hot poker them or get them comfortable to go ahead and do this, what would you tell them to get them to take that leap of faith that you took?
Wow, that's a great question, because not everybody likes to dig into the weeds like I do. But, you know, I think if you look at it from the standpoint of, you know, what's a good investment that I can make, even if it cost me an hour or so of my time to come up with something that may save me over the course of a week or two weeks or a month, several hours. I think that's the kind of key there. It, it's about how can I take these new tools that are really kind of incredible, just blow my mind, and use it for good? And not, you know, I'm not creating, you know, the the dancing cat videos with it, right? I'm trying to figure out how to make it work so it saves time and makes us as a team more productive, so we can spend our hours doing the things that matter for our clients. And that's kind of how I looked at looked at my investment in this. Yeah.
Well, uh, the other big questions that we're getting in, we can't answer all of these individually on this webinar, uh, are really about broker dealers, RIA settings, organizations that are allowing to you to use, uh, MCP. Because what, what we're hearing is not a lot of the larger firms really want you doing this. Uh, so this is really, really great if you're an RIA, uh, and you're really monitoring yourself, but you still need to make sure that you're checking with your compliance organizations to see how they feel with that.
Um, all right, Mark. I'm, I'm gonna, I'm gonna tee you up for the question that I asked you at the end of our podcast. Uh, but what, what did we not cover? And Neil, I'm gonna ask you the same question. What did we not cover so far today that you want to make sure the audience knows about?
Does Neil have to go get to go first or do I? I'm, I'm up first.
Well, yeah, I'll have you go first.
A hard one, Neil. Um, yeah, I'm, I'm fine with that. Like, there's a lot. I'm just getting to the chat. So, and we will try, I think, in the remaining time to just rip through and frankly answer as many of those as we can.
Um, I think a big one, like most of the things I'm seeing are really around kind of like data, right? And where is data and where is it going? And so, I'll just kind of echo something and and try to simplify it. I, you know, one of my challenges is always like, how do we simplify the data story down enough? Um, you really want to be thoughtful about what systems have access to your PII and not. Um, if you have PII in, um, Zox, uh, depending on your settings, you'll, you know, you'll probably know if you do or don't because it'll get redacted if you don't. Um, that data can go through the MCP interface right now. So, you would want to have Claude, ChatGPT set up in a way that, you know, it is not storing and retaining that PII data. And frankly, if you're, if you're associated with a broker dealer, an independent broker dealer, you need to make sure they've approved it, because that's exactly what they do in that, you know, in that review and approval process. If you don't have that, we do have something coming out very shortly that I won't get into specific timelines on, but I'll just say very shortly that allows to, um, to essentially help with that at the, at the layer, if you will, at the interface between Zox and that system. So that if Zox, I'll say, is more trusted in that system, you can do that.
Um, I would, you know, I'd love there was a quick answer for this though. I think Matt, you put a note in there. If people have questions, you can always get a hold of our team and kind of walk through the specifics. There are a lot of details around like, you know, we see a lot of different fields in Orion, but there's like reports and certain things we just can't see. So if we really understand your scenario, we can probably tell you very clearly, can you do that directly in Zox? Is it better to use the MCP interface and go through it? And depending on, you know, your broker dealer and everything else, like what options do you have there? Like, is it co-pilot? Is it something else? Are you in kind of a review and waiting process? So I would just tell people to kind of reach out to us and we're happy to to kind of go into their scenario in more detail as well.
Yeah, we've got a heck of a team, everybody, who is, uh, uh, that's what they're doing, uh, all day, every day, is is meeting with all of you and answering those questions, uh, because they are the people who can help you use. One of the things that we really want is for you to stop just using 10% of the product and really start using a lot more of the product, because the, the dev team's working really hard on all these really cool features that we want you all to start using.
All right, Neil. So, I'm going to punt that question to you, uh, which is, what, what should I have asked you that I didn't?
I, I don't know that there's anything that I would say other than just, you know, continue to be, you know, be curious and and not be afraid of technology, right? You know, it's, it's a big part of our daily lives. And the more we embrace it, the better I think we are at using our time wisely. And the tools that I've seen in my 30 years have have made that very true. And this is just another one. And I think that the big question is, you know, why not use, you know, figure out how AI can help you and instead of being afraid of it.
Yeah, change is hard and scary, and and we understand that in an industry where you don't want to be doing a lot of changes because it can make your clients upset and concerned. This is still one of those things, and I'm going to briefly tell a quick story. Uh, we, we've had, uh, many of our clients, uh, win business, uh, from other advisors, or get more referrals with the speed to service. And that might not be technically what everybody's calling it right now, but, but a client comes in and sits down and meets with Neil, and Neil has this unbelievable prep done. He's also got this on the back end, too, which we didn't really talk a lot about today, which is all the follow-up stuff and all the other things. When you can deliver an unbelievable summary, tasks in order to make sure that they're going to do the stuff, because Neil was just talking about that, the stuff that they need to do, the stuff that you need to do. And if you, if they have other advisors, which I don't remember the number currently, the last time I checked it was like 65% of clients have more than one advisor, if all of the sudden your speed to getting their their results because you're leaning into technology is exponentially faster than their other advisors, you're going to win that business. And I just think that's really important to keep in mind. We are business owners. We are business people. And business is a really important component.
Now, Mark previewed this a little bit, and I'm going to preview it a little bit more. Uh, on June 24th, I need all of you to mark your calendar because we're going to launch something new. All right. In my 20 years, when I saw, I got demoed this on Friday. This, how quickly this is, you know, they, they don't show me stuff because they know I'm going to talk about it, which, by the way, is really smart for them to do and not show me too much under the hood. But I was demoed this on this past Friday, and I have to say that this solution is one of the biggest shifts that are going to happen in your practice to make your overall life better. Zox is now going to be able to help you have better service, grow and build your business, and generate more organic growth. And I'm not overselling that because I'm not even scratching the freaking surface, uh, or surface on what's going to happen. So, please, please, please, uh, we're going to drop the link in the chat here in just a minute to go ahead and register for the webinar on the 24th. And if you are not a current Zox user, please make sure that you, you go ahead and scan this real quick, and, uh, we'll be more than happy to take you through a demo on how all of this works.
I want to thank all of you for taking time out of your day to learn more about this technology. This technology changed Neil's practice, and we hear it all the time at conferences and in just meetings with our clients that Zox fundamentally didn't just change their practice, but helped them live a better life. And when you have a better life and a better practice, isn't that what you're all doing this for? I think that's an important thing to leave everybody on. So, uh, if you're not following all of us, Mark, uh, Neil, or myself, make sure you connect with us on social media, and we will see you on June 24th for our next webinar. Thanks, everybody.