Insellerate

Transcription:

Josh Friend (00:08):
Well, thank you everyone for having me here. Appreciate the time. We're going to show you guys a couple of things today. First, just real quick, just to give you an idea for those who know Inseller and who we've been: we've always had a marketing CRM platform, integrated telephony, and realtor tools. We've really sat in the hub of lenders for quite some time, being able to manage large amounts of data. We originally came out in the consumer direct space, which gave us access to huge amounts of customer data. So with all that, we started to realize there's going to be a lot of opportunities for AI within the data that we have. We actually started our AI journey in 2018. If any of you guys were at the Ellie Mae Show before it was ICE in San Francisco, you might've come by our booth.

(00:51):
You would've actually been able to work with our AI bot that could talk to you through text messaging, take a mortgage application, get you qualified to buy a home, and tell you what you could afford. Believe it or not, in 2018, people weren't really ready for it. It's actually kind of funny; people came up and said, "Either AI is never going to talk to my customer," or "AI is not a real thing, so I'm never going to need this." Seven years later, it's really come a long way. We realized that wasn't the product that we needed, so what we went into next was understanding that we have large amounts of call data and interaction with customers. We figured out how we can help our lenders with that. So, we started using AI to understand sentiment analysis and calls.

(01:26):
It was really neat to tell you what kind of customer you had, but it didn't really move the needle. So we went into some additional things with machine learning, tried to learn some predictability, and understand what loans would close and how they would work. We took machine learning and then we took ChatGPT in 2023 and we started working on the process with CrossCountry Mortgage as our first customer. We rolled out the Athena platform on January 1st, 2024. In 2024, we reviewed 50,000 hours of calls. We've funded over a million loans on our platform, so we have a lot of data to be able to really understand what's going on. Athena's really grown.

(02:01):
Athena has turned into much more than just analyzing calls. To give you some high-level numbers: now we've analyzed 180,000 hours of phone calls. We can predict for CrossCountry Mortgage with an 89.4% accuracy rate, when they talk to a customer, whether or not they're going to fund a loan with that customer. Think about that. We've had over 800,000 calls we've analyzed and we have a lead score on that. If you guys look at the right, it's very powerful and we can go into it later, but I know I don't have a lot of time because this is going to be a demo. Now that Athena has listened to 180,000 hours of calls and had 800,000 calls it's listened to, Athena has a few things to say. Athena started talking to our customers about four months ago: scheduling calls, answering calls, reaching out to internet leads, following up with customers, and getting information and contact data.

(02:51):
And we've all seen that, right? We've all seen these bots that can talk to your customer and schedule a call or follow-up call. What we wanted to do is really go beyond that. Today we're going to show you some of our agentic AI with Athena. This is new; this is in beta. This won't be ready for all you guys until probably the middle of Q1, but it is being used now. We're going to stretch what Athena can actually do besides just making calls. Let's go ahead and open up the system. We're going to go into our CRM and we're going to input some data real-time live to show this is real, and we're going to have some interactions with Athena. Now, if this was used in our platform, you don't have to use our CRM. This can interact with any CRM, any database, or any LOS that's out there, but I'm just going to give you an example of how this works with us. All right. So I'm going to need a volunteer in the crowd. Anyone? Will? Alright.

(03:46):
Give me your phone number, Will, we're going to go ahead and put it in here. I put Julian Hebron's name in as the processor in this file just because I thought it'd be easier to do that than try to do it live. So what we have in here are some dates and some information about this customer, and I'm going to go ahead and give a call. So let's see.

Athena (A.I.) (04:32):
CrossCountry Mortgage. This call is being monitored and recorded for quality and training purposes. This is Athena speaking. Who do I have the pleasure of speaking with?

Josh Friend (04:44):
Hello, this is Josh.

Athena (A.I.) (04:48):
Please hold one moment while I pull up your account. Great. I found your loan file. It looks like you are looking to purchase. How may I help you today?

Josh Friend (05:01):
Hey, I was wondering, I know I have an appraisal coming up, but I'm not sure when I scheduled it. Can you tell me when my appraisal is scheduled?

Athena (A.I.) (05:12):
Your appraisal is scheduled for September 25th, 2025. How else can I assist?

Josh Friend (05:18):
I was wanting to know, can I reschedule that maybe for October 10th, 2025?

Athena (A.I.) (05:25):
Rescheduling to October 10th isn't possible. It's after the contingency deadline. Would October 3rd work?

Josh Friend (05:33):
Why can't I schedule it after the contingency guideline?

Athena (A.I.) (05:38):
The appraisal must be completed before the contingency deadline, which is October 6th, 2025. This is the deadline in your contract to finalize your mortgage approval. By that date, you need to either confirm your financing is secure or back out of the purchase without losing your deposit. Scheduling the appraisal after this date could jeopardize your ability to meet this critical deadline. Would you like me to help reschedule it to an earlier date?

Josh Friend (06:09):
Yes. Can we do October 3rd?

Athena (A.I.) (06:15):
Please give me one moment while I text your listing agent, Will Lowe, to verify the new schedule.

Josh Friend (06:22):
Text Will.

Athena (A.I.) (06:26):
Yes, I want to text your listing agent, Will Lowe, to confirm the reschedule for October 3rd. I'll update you as soon as I hear back.

Josh Friend (06:39):
Great. Do you know when my termite inspection is?

Athena (A.I.) (06:44):
It looks like there's no date listed for your termite inspection in the file. Let me email your loan processor to get more details. One moment please.

Josh Friend (06:55):
Who is my loan processor?

Athena (A.I.) (07:01):
Great. I just sent the email. Is there anything else I can help you out with today?

Josh Friend (07:06):
Who is my loan processor?

Athena (A.I.) (07:10):
Your loan processor is Julian Hebron.

Josh Friend (07:14):
Okay, thank you very much. Have a great day. So that's the power of using more of an agentic AI system. Will got a text; if we look in Julian's email, he is going to get an email saying, "Can I reschedule?" or "When's the termite inspection scheduled for Josh Friend?" What we wanted to really show you is just going beyond answering a call saying "When can I call you back?" to how this can actually help you do work. We have a lot of different stuff in the platform, a lot of other AI stuff. Hey, real quickly, we always have a giveaway; scan this. There's three questions. Whoever answers those questions right goes into a raffle and we're going to send you a gift card for 500 bucks. These are three questions that go over some of the stats from Athena.

(07:57):
I think you guys can scan it and go to the next slide. I also want to encourage you all on October 9th, we're having a webinar. We're actually going to be showing some other AI stuff we've built in the platform, but not just what we've built. We have Neil Sahota; he's authored multiple books and he is a United Nations subject matter expert in AI and emerging technologies. He's quite the industry leader, an advisor to Insellerate, and a good friend of mine. He's going to be showing some best practices for mortgage lenders on how to use AI. We're also going to be showing people how to use R.E.P. if you guys have ever seen that, and how to use it within the mortgage space. We actually have R.E.P. built into our code base. It's pretty cool; if you haven't seen that in action, I'd suggest you guys take a look.