Building Claude Into Your Title Company: A Real AI Rollout Plan | Ep 97

In a world where change is the only constant, Mo Choumil stands at the forefront, guiding title professionals to not just grow their businesses, but to master the art of innovation. With every episode, you're handed the keys to unlock unparalleled growth and stay ahead of the curve. Get ready for a transformative journey. Hello everyone, and welcome to the Title Agents Podcast. I am your host, Mo Choumil, CEO of Alltech National Title.

Today I'll have our regular guest, Hope Araviani, who is COO of Alltech National. Hello, Hope. Hi, everybody. This is one of those AI Corner episodes. We're going to share with you what we're doing here at Alltech National and ATG Title.

We recently, about a month ago, maybe less, rolled out— I'm a big believer and Hope is a big believer as well in Claude. It's Anthropic Claude. There's a— we talked about different LLMs, ChatGPT, Claude, Gemini, Grok, Perplexity. There's so many of them, but Claude is definitely so far is the brightest, smartest, and the one we're going with moving forward. And recently we've rolled out Claude team.

It's an enterprise platform. It's closed environments for our company. So Hope, um, wanna talk about the AI initiative, uh, and rolling Claude team to, to come to you? Yeah. Yeah.

And it all started with a meeting with our leadership team on how we were gonna implement what we wanted to do through Claude, right? Through AI. So first we, we had the meeting, we chose the platform that we wanted to use. We got the enterprise closed environment and then we first created this shared project and that's called the AI rollout, how we're gonna implement it, the steps, the deliverables. And then I was able to take all those deliverables over the next 6 weeks and auto-import them into my calendar so I could hold myself accountable.

And then from there, we just broke it into chunks. Just like Mo says, how you eat an elephant, you break it into chunks because if you looked at it, As a huge, you know, everything on one page, it gets overwhelming. People lose interest and then they go on to something else. So the way we were able to succeed so far is just break it into chunks. And that's what we did.

We broke it into chunks, or AKA our projects, our shared environments. So we are currently still building this, still refining and retuning, but this is the rollout plan. I've got my deliverables, and then every week we just consistently do training touchpoints. Uh, maintain those, uh, communications with all your departments and then tell them the vision. If you don't tell the people the why or the vision or what you're planning, they're not going to buy in.

They're not going to have interest. To all of you that are either managers or company owners, the proper and the only way to really roll out AI initiatives is start from the top. I've been talking about AI for the last 3 years and I'm a big advocate and The positive power of AI and I'm very blessed and lucky to have an amazing COO, Hope Oraviani, kind of just really sucked all that energy in and another big advocate. So it has to start from the top. So I started first, I really got my feet deeply, deeply in the weeds and I've been coding.

I'm a coder now. I just joke around. I'm becoming a nerd. Yeah. Nothing against all the nerds out there.

We love you. But I'm becoming one of you. So, yeah, and literally kind of teaching gradually, it's a massive, massive fire hose and it's just so much information, so much changes, so many updates. And it started off from the top. Like, we rolled it out to our leadership team, regional managers.

Next is the sales team and then the operation team. That's so true. You got to start with the leadership because you got to get their buy-in first because then they will waterfall it better. to their regions, their departments, and their teams. That, that way it's not just on me and Mo who are huge advocates and energized about it.

We're getting other people energized so that they can energize their teams and it doesn't seem as overwhelming. So we did our rollout plan, we set our outline, right? We started with the leadership. That's great. Next step is to build the projects and the departments.

So what is a project? Yeah, so a project is a shared environment that you can invite certain departments Feed it. It's kind of like the AI brain, if you will. Feed it your core values, your context, if you will. Maybe that department's SOPs, processes, your company, the way you guys communicate and conversate.

So that's what we did is we first started each project. I titled it, and then I started feeding the brain with all the SOPs that I've ever had over the 15 years I've been here. I refined them, I finalized them, I updated them, and then I put those final versions into each project. So as you can see— So what do you, what do you have here? Your financial accounting.

So— Yep. So let's run through them. So I've got, you know, the, the outline, of course, AI company rollout. And then from there we have finance and accounting. We've got title examinations for our title officers, post-closing and funding, HR and people operations.

Compliance and legal, processing or escrow officers, sales and business development, marketing and content, and leadership and strategy. And, and we may have more, we may build, we may refine, retune, like we always say. But from here, this is kind of our framework. We've already added like leadership and strategy. We've, if you scroll down, as you can see, we have all of our brain and it's not even completely, you know, it's 5% capacity used.

So you can kind of see. Yeah. You're, yeah, you're feeding the brain, you're giving it the information, and then from there you're able to build artifacts, build projects, share it with other team departments, and then they can kind of utilize it and build projects too. So it's a whole project we're building out. We're just 30 days in, but you can see the framework and we're taking, and I, I wanna be, you know, transparent.

We use Microsoft OneDrive right now, and Mo, I know does not like it. It's just not friendly. So we're using that OneDrive and just making it AI native. We're taking all that information that we've been saving for years and then putting it in here so that if I have a question on a specific domain or department, I just go to that domain and I ask the question, you know, how do I, what's the exception for Wisconsin and transfer on death? And you can, you can ask it because we're feeding the brain with the underwriter manuals, the information, the TitleGPT.

So. Yeah. You're feeding each of those projects with the information. And then from there, your team should be able to log in and just ask AI questions and it'll be able to answer for them. So that's where we are right now.

Yeah, I know. We've forgot to mention the first step when setting up the whole Claude team. Yeah, you start off with the, it's called the master memory file or constitution or whatever you want to name it. Yeah. And literally we put up our company history, our strategies, our goals, our values, fed as much information about key, the leadership team, and it's all this massive 3 or 4 page document with as much information as possible about a company so it has a context.

So we also has the RESPA guidelines, RESPA rules. So when people are asking questions, trying to do something, it always has that RESPA rule in mind. So it really kind of has some massive, massive guardrails so people don't go kind of outside the box. Yeah. And from there, when you create that master file, then you create projects, but you really have to start from the top, that master file, the master memory file, just it.

And the projects we have here, the files that Hope showed, they're nothing but a memory file. So it's called MD or Markdown file. It's a, basically it's a text document instead of a Word document. You use a text document. It's just terms you need to get used to, MD files and markdown file.

And it just, if you go to examination. Yeah. So it just has a bunch of documents, has a lot of information. That's how, what it uses on top of the language model knowledge base of title insurance, escrow rules, regulation. It has a kind of this universal knowledge that it learned on, like how LLMs were built, then now adds our additional kind of, um, customs and rules and guidelines and underwriters' guidelines.

So just kind of, it's really very specific to that project. Mo, if I was an outside company, a COO, how did you start that index file? Because that's what I'd want to know. Because I, did you just open up a chat box one day and feed it the information that gave you that index file? Like, how did you start that process?

One of the prior episodes on the AI Corner, we talked about prompting. Mm-hmm. And, uh, it's literally, it's all about context. Like, it's the mass— the most important thing, how you prompt it. Like, just to— they say the garbage in, garbage out.

If you ask kind of a vague question or vague something like that, it just gives you vague information. That's why people kind of get discouraged. They're like, oh, it's useless. Wow. It's literally like now if you want to email me, I can literally send you a prompt, like direct prompt.

It's mo@alltechnational.com or Hope. At alltechnational.com. We can literally go, we should literally show you exactly how to go through that process, which is very, very key. It's foundational if you want to use AI properly. And the beauty of it, if you switch platform, let's say we're now, we're big on Claude.

And I, what I always say, like the next best thing comes up. Hopefully. I don't know, Mo. But so the having those master files, those are transportable. You can transport that information from one platform to another.

And I personally, as I tell you, Hope, I'm using all of them, like Gemini, Grok, OpenAI, Flexity, and they're all based on encryption. Like, I have a personal account, but this is the business account. I use all of them for different reasons. They all have my master file. When I say, okay, this is my— it knows me exactly off the bat, you know, whenever you sign up, it knows me exactly to my core.

Yeah. Yep. And so every project that I start, if I start a new project today, the first file that I'm uploading is that master context file that Mo and I have built that gives you the core values, the context of your company, what you do, the why, the purpose. Yeah. And, uh, the Ask Alltech button, uh, to the left has a knowledge base by the company just to, for employees and team members to ask different questions about HR, about, um, All company knowledge.

Yeah. It's a, it's a cool, we ask whatever title company you are, a great, it's a, I don't know if you, uh, I know, um, the, the, the big underwriters, the big companies have this kind of knowledge base or intranet inside their portals, whatever. So this is kind of really democratizing, having like your own intranet with a closed environment. It's a, so employees and team members can find information a lot quicker. And more efficiently instead of trying to like get so many pages to find, okay, what's my pay?

What's my whatever that is. That, and that's always been the problem for a leader is everybody comes to me. I mean, we have, you know, 80 employees. Everybody comes to me and asking questions. What's our PTO calendar?

What, you know, what date are we off for the holiday? It's just simple questions that are just flooding my inbox. They could easily just go to Ask Alltech because the information's there and then they'll get a quicker response. Versus having to wait for me, who, you know, I'm doing several other things. I mean, I could ask them till later in the day.

So this is going to cut time, you know, communication. Uh, anything you want to add in this episode, say a high level, uh, as far as Claude implementation? No, I just start now. I mean, AI is so dynamic and changes faster than I can think. And it's, you can't put it off.

Oh, I'll get to it in the, in the spring, or I'll get to it in the summer. You have to start now. You're already behind, right? So That don't wait to spend an hour or so setting up that outline like how we did. We had a company meeting, we got together and we set out the framework, and then you just need that implementer to kind of put it into motion.

So I think that's what you just have to do is start now. The best analogy, hopefully this will come get through to you if you're kind of hesitant or not sure, would you not use the internet today? You can't live without internet. Even as old as you are, Google. There is literally, and I mean, I mean this from the bottom of my heart, it's exact— it'll be like a year from now, 2 years from now, you cannot live without using AI across the board.

So either jump on the bandwagon and adopt it early and, and succeed, or be left behind. Literally, you'll be left behind. Yeah. Yeah. Raise your profile.

Well, well, uh, thank you, Hope. And, um, until, uh, next episode, we're gonna dig deeper into Claude and, uh, some of the dashboards, workflows Hope is using. Uh, until next time, have a wonderful day. Bye. In a world where change is the only constant, Mo Choumil stands at the forefront, guiding title professionals to not just grow their businesses, but to master the art of innovation.

With every episode, you're handed the keys to unlock unparalleled growth and stay ahead of the curve. Get ready for a