- 4 months ago
- #mortgagetech
- #loanorigination
- #fintech
- #vestalos
Vesta is the next-generation mortgage LOS that reduces lenders’ operational cost by “inverting control,” guiding users through the loan. With data-driven tasks, validations, and native automations, Vesta eliminates manual work — enabling faster closing times, higher loan quality, and greater efficiency.
#MortgageTech #LoanOrigination #FinTech #VestaLOS
#MortgageTech #LoanOrigination #FinTech #VestaLOS
Category
🤖
TechTranscript
00:00Hi there and welcome to HousingWire's Demo Day on Demand. I'm Allison LaForgia, Managing Editor of HousingWire's Content Studio, and I'm glad you're joining us.
00:09Demo Day is where we spotlight the most innovative technology companies in housing and mortgage, giving you, our audience, a front row seat to real product demos from the teams building the tools that move our industry forward.
00:20In this session, we're featuring Vesta, a company that's solving real-world challenges with smart, scalable solutions.
00:27Stick around after the demo and we'll dive into a short conversation where I'll ask the questions that'll help you connect the dots between their tech and your business needs.
00:37In this Vesta demo, Mike will be taking us through Vesta LOS.
00:41Vesta is the next generation mortgage LOS that reduces lenders' operational costs by inverting control, guiding users through the loan.
00:49With data-driven tasks, validations, and native automations, Vesta eliminates manual work, enabling faster closing times, higher loan quality, and greater efficiency.
01:01Mike will show us how the industry's most tech-forward lenders are preparing for the future of mortgage origination.
01:07Mike, take us through the demo.
01:09Awesome. Thanks so much, Allison.
01:11So today we're going to walk you through a little bit of Vesta.
01:13Of course, there's an entire LOS to kind of explore here, so people should always feel free to reach out to see more.
01:20But in particular, we're going to take a focus on how we at Vesta think about workflow, making that task-based, and how that drives efficiencies for lenders.
01:26So we'll start with kind of a dashboard.
01:28We'll take you through some of the core concepts of objectives and tasks and how Vesta really chops up the loan and guides the user through what they need to do next.
01:35And then we'll actually go into a queue, do a bunch of those tasks, see some of the experiences that we at Vesta have built around those, and how automation can make those even more efficient.
01:45So to start, we'll just explain some of the basic concepts.
01:48We're looking at a dashboard here at Vesta.
01:50This is a pretty basic, you know, we've got loans and application, and we've got a bunch of pipeline views that we've pulled into this dashboard.
01:56And then we've also got a bunch of work pipeline views as well.
02:00So if I want to take a look at my 60 loans and application, for example, I can always drill down, take a look at my actual loan pipeline, and see something that I think probably everyone who's used in LOS is going to be pretty used to, a list of a bunch of loans, a bunch of columns and filters that you can set up.
02:16However, in Vesta, the loan is actually not the atomic unit of work.
02:19So you can keep track of your loans as a manager, always a really important thing to do.
02:22But we're actually going to have the system automatically compute a bunch of work that has to be done on each loan, and that you manage the work directly outside of the loan.
02:30So for example, I might have a processing management dashboard, and I can see here that I've got a whole bunch of objectives assigned to different people.
02:37Mike has 46 objectives to do, Graham has 57 objectives to do, et cetera.
02:42If I go and drill down into that, I'm actually going to get a work pipeline, which, instead of showing me loans, is going to show me actual pieces of work that people have assigned to them.
02:50So for example, Graham, in his 56 objectives, needs to actually, you know, collect documents from the borrower on loan 785, or he needs to pre-approve and disclose a whole bunch of loans, or he needs to review some borrower docs on loan 713.
03:05And so you can actually see all of the work that your people are assigned to do and manage that, both as a manager and, of course, ICs can take a look at this view as well and see only their own work.
03:14You actually can even do that work directly from the pipeline.
03:17So if I go and I take, you know, my own work, for example, as Mike, and I say, hey, I need to go ahead and pre-approve and disclose loan 677, I can actually open that objective up, which is going to be a group of tasks, and go and execute those tasks right away, right from this loan pipeline, or work pipeline.
03:34So I can hit start, and it'll just pull me into the loan.
03:36It'll pull me to the checklist and the things that I need to start doing.
03:39I can go ahead and just kind of start working on the loan.
03:41So, for example, if the first thing I need to do is pull credit, I can really quickly go ahead and set this up, run my credit pull, et cetera.
03:49And so really the core idea of Vesta is the system will figure out what work needs to get done, and then it'll kind of guide the user to do that work.
03:56They can take a look at a prioritized list from the work pipeline and kind of go from there.
04:00So I've pulled credit.
04:01You know, maybe I finished that, and I'm going to go ahead and say I've done the item, and it's going to take me to the next place I need to do, for example, to browse loan products.
04:09So that's kind of the core idea of an objective.
04:12Lots of individual contributors work out of a work pipeline, but we've actually taken things even one step further, and we've got this idea of a queue, which I'm going to show you now.
04:21So if I go over here to the left-hand side, and I say I want to enter my work queue, the idea behind the queue is that it's actually going to be extremely prescriptive with what a user needs to do.
04:29So it's going to say you're going to do exactly one thing at a time, and we're going to give you the one highest priority objective in the entire ecosystem that is most, you know, relevant.
04:39to your skill set, and that is most urgent to do.
04:41And as a user, we actually have customers who say, my ops people log in every day, and all they have to look at is this queue.
04:47It keeps them focused.
04:48It makes them more efficient, and it allows you as the executive to kind of decide exactly what it is you want your people working on at any given point in time.
04:55So right now, my prioritization function says, hey, Mike, the most important thing you can do with the skills that are configured in the system for you and with all of the loans in the pipeline is review the borrower docs, the docs that Andy provided on loan 790.
05:07So we'll go ahead and we'll start doing that.
05:10This is a place where there's a lot of opportunity for automation and AI to make things much more efficient.
05:15So, of course, we've got the docs.
05:17We're going to do a pretty basic review here.
05:19We're going to say, hey, is this an acceptable pay stub?
05:21Is this not an acceptable pay stub?
05:23If we say yes, we'll notice that we've actually been able to auto extract the whole bunch of the data.
05:27We are using kind of, you know, OpenAI or Google Gemini under the hood to do this.
05:34And so what this actually allows you to do is say we've got a whole bunch of different documents that we can, different document types that we can pull in and the LLM will kind of figure out what to do.
05:46So here we'll pretend this pay stub is not eligible because we want a Tesla pay stub and this says TESTA.
05:51So this doesn't seem like the right employer.
05:55And we actually can then have the system say, hey, so as a user, do you want to re-request this original document, reopen that condition?
06:02And we'll say, yes, we'll reopen the upload pay stubs task for the borrower.
06:05And we'll say, you previously provided a TESTA pay stub.
06:11Please provide 30 days of pay stubs for Tesla.
06:15We can submit that and it'll immediately take us to the next task in our objective, which is to review this W2.
06:20Maybe at this point we're like, oh, maybe the TESTA docs are fine.
06:23It's really fast.
06:24You can see for the user to go ahead and say, yes, this time the AI actually managed to get all of the data off the doc.
06:29And we'll go ahead and submit that and it'll take us again to the next doc.
06:33And so you can see how quickly a user is able to kind of review these docs, confirm that everything is right, reject the document if it's not the right document, and then accept the document and populate the fields aided by this AI extraction.
06:47In this case, for example, the AI missed the purchase contingency date, probably because it says something like 21 days after the contract date and doesn't actually say what that date is.
06:57So we'll go ahead and we'll just put in 21 days after the contract date as the contingency here.
07:02And so we just have to fill out the one field.
07:04We can check the rest of the fields.
07:05We'll update all of that data in the right place on the loan and we'll continue through.
07:10And so really accelerating, you know, a processor's ability or an ops person's ability to go ahead and review these docs.
07:16So I'll skim through the rest of these now, you know, get through these six docs in just a couple minutes.
07:21And we can extract, as you can see, all kinds of different docs here.
07:28And this is natively built in the LOS.
07:30You can see it's really nicely integrated into the workflow.
07:32And that's actually a big thing we've seen when it comes to document intelligence adoption is really making sure it's integrated well in the workflow.
07:38So now that we've completed that, you can see that objective actually automatically completed.
07:42We did our six tasks.
07:44We've completed two objectives today now.
07:46And it's going to say, hey, here's the next thing to do, Mike.
07:48Now you've got to go review DTI on a totally different loan on 676.
07:52But this really allows you to facilitate, you know, your users working on always the next most important thing.
07:59So now we'll say, hey, we're going to get a checklist to review a credit report and some liabilities.
08:03So we're going to take a look at the liabilities.
08:05We're going to check for credit report characteristics.
08:07When Vesta pulls credit, we'll automatically pull in things like the inquiries and let you quick filter them.
08:11We'll pull in the late payments.
08:13So making it really simple and easy to kind of do that task.
08:17And then the next task is actually going to be to calculate the income.
08:20And so in order to calculate income, we'll go ahead and we'll try out our income calculator.
08:24So just another way in which we've built workflow and UX directly into the LOS to facilitate the completion of these operational tasks.
08:30We'll do this income calculator and we've got a couple incomes on this loan to calculate.
08:35On the employment side, you know, we've got a pretty complicated calculator.
08:39You'll remember we actually used document intelligence to extract all that data from the documents.
08:43That data is going to get prefilled into the calculator so the user has access to the documents if they want, but they don't have to necessarily fill it out all over again.
08:50We're using W-2s and we've got, you know, a couple W-2s with box 5 extracted.
08:55And then we'll just flip this over, make sure this looks right.
08:59And then the system will go ahead and give us some ideas for the income calculation.
09:03Now, as an underwriter, I might always want to be able to override this and say, hey, I actually want to use the year to date amount for some reason.
09:09And I can, you know, be forced to leave some additional comments explaining why I picked something that the system didn't suggest to me.
09:16But underwriters do love that.
09:17And I can say, hey, my variable pay, it looks like I've got about $7,000 a month, but there's a big decrease between 24 and 23 average of variable pay.
09:27And we've got year to date.
09:28And so we're going to go ahead and ask for an additional condition.
09:31Maybe we're going to say from the borrower, we're going to ask them for not just their 24 and 23 W-2s.
09:37We'll also get the 22 W-2 just because, you know, we're a little bit nervous here.
09:40And so we can seamlessly kind of add an additional condition from this income calculator, take a look at how that trending analysis works out, finish the calculation, and then quickly apply, you know, calculate the Social Security income as well.
09:56We'll see here that we don't have a tax return, so we'll just take the conventional guideline and gross up 15% of it by 25%.
10:02And we'll wrap that up.
10:04We'll apply the calculated income.
10:07And we'll say we've finished that item in the task.
10:11Since we've changed the income so much, the DTI moved a lot.
10:14It's going to prompt us to rerun AUS.
10:16Vesta is always going to be dynamically in real time recalculating everything on the loan, figuring out where that creates discrepancies, where that creates problems, and, you know, making sure that the data is consistent and that you have everything you need at all times to make sure that loan is going to get closed quickly and efficiently and you won't run into something later.
10:34So that's going to create a validation.
10:35It's actually going to stop me from being able to submit this task.
10:38And so when I go and submit that task, it's going to say, hey, you need to, you saw that warning, which said, hey, someone's going to have to go and look at this before you can move that loan forward to run AUS.
10:48Finally, you know, people often ask, I'm sitting here in a queue.
10:52I've got a bunch of work.
10:53What if I can't do this?
10:54What if in this case, you know, it's asking me to get pay subs from the borrower and I can't talk to the borrower.
10:58I'm just an ops person.
11:00Someone misconfigured something.
11:01We, of course, have a whole bunch of workflow around escalating to your manager, around blocking for a certain period of time that can be configured as well.
11:08So this is really just a quick tour into how we think about doing work here at Vesta.
11:13You've got these task experiences that are really designed to facilitate that efficiency.
11:17And then you've got both a work pipeline and a queue to help your operations people decide what they've got to work on next.
11:21There's tons of magic and tons of configuration that goes into how all this workflow gets spun up automatically, how it gets assigned to various people, and, you know, how all of this kind of gets stitched together to complete a loan.
11:33But it's, I think, a very different way of taking on originating a mortgage, and our customers have seen a lot of efficiency and success doing it.
11:40So hopefully this was a good overview to how we think about workflow here at Vesta.
11:44And if people want to learn more, they should always feel free to reach out or ask some questions.
11:48Mike, thank you so much for taking us through Vesta LOS and showing us what it's bringing to the table and how it's eliminating manual work, enabling greater efficiency.
11:58Let's dig a little bit more into the platform and jump into some questions.
12:05So the first question that I have for you is, where are lenders on Vesta getting the biggest lift in efficiency from automation?
12:14Yeah, I will say there are probably two things.
12:16Actually, the biggest one, especially at lenders that are more midsize, does tend to be that they, because you can use task-based workflow and write a bunch of rules to enforce quality earlier in the process, a lot of the efficiency actually comes from underwriting and downstream efficiency, where those people are seeing the loan fewer times.
12:34They have to do less work on each loan, you know, a lot of it has been automatically conditioned, which helps, but even more importantly, it means a bunch of those conditions are fulfilled before they see the loan the first time.
12:44And so the more you can kind of shift some of the thinking into the machine and accordingly shift it upfront, we're seeing a lot of good, you know, underwriting efficiency that our customers are telling us about, because they've been able to basically do fewer touches on each loan.
12:58The other big advantage, of course, is you can probably see here, automatically figuring out what work to be done, needs to be done is a big advantage, and then automatically doing some of that work, whether that be things like the document AI, or we have similarly like automated service orders, which we didn't get to show here, that also can drive meaningful efficiency as well.
13:19And let's talk a little bit about AI, since it's such a hot topic right now in the industry.
13:25How is Vesta leveraging AI today, and how do you see that evolving over the next 12 to 18 months?
13:36Yeah, I mean, today, I think the doc AI is really the big focus for us.
13:40It's kind of the obvious place in the mortgage industry where I think a lot of people see an opportunity to say, you know, we can take something that can structure this unstructured data, and there's so much unstructured data coming in in the form of docs.
13:49Over time, I think lenders can probably also imagine, once you've built kind of this really nice experience that allows, you know, more entry-level employees to just do exactly what the system is telling them, you can absolutely overlay an AI agent on top of it and tell it to, you know, do that work.
14:05I think I've actually, some people might have noticed, eagle-eyed viewers might have noticed in the demo, I had, like, a user called AI Agent Test, which is, also had, like, eight objectives assigned to it or something like that.
14:17So I have played with, you know, you make a persona for some AI agent, you give that login, actually, to, we use an AI coding agent here at Vesta, and I was like, we'll just let the coding agent try doing some tasks.
14:26And it actually did, you know, it wasn't tuned for it, it wasn't outstanding, but I certainly think that there will be an opportunity to have an agent just sitting in a queue that looks like this, churning out tasks as well.
14:36And it's kind of like, you know, getting an entry-level employee who can do some of the easier stuff, like reviewing the docs or placing service orders.
14:44Sounds like a really interesting potential future capability.
14:47I hope so.
14:49I think there's a lot of promise there, and I think that the most important thing is, really, you need the system to be making the hard decisions, like, what do you work on that's got to be made deterministically?
14:58What is the credit decision, of course, has to be made deterministically?
15:01But once you've actually figured out what work needs to get done, anything that, for example, you'd be comfortable offshoring, I think you probably should be comfortable, you know, sending to an AI agent as well.
15:09And I think it's just much easier to send something to an AI agent than offshore, especially for some of these lenders who, you know, don't have the scale to set up a whole offshore operation.
15:18Absolutely.
15:19And to what extent can Vesta be customized?
15:23What freedom does the lender have to develop their own bespoke needs, and how does that get done?
15:31Yeah.
15:31So everything that we saw here, for example, in the workflow, that is all customizable.
15:36We have a point of view, which is the lender has to have exactly the process that they want, the way that it's routed, who's responsible for doing it, the way the words show up on the screen and tell people what to do.
15:46In order for this to really get adopted, because, you know, while everyone is kind of originating to a similar set of guidelines, and that means we can give them something out of the box from a rules standpoint, people still have unique operations.
15:57They have a unique way that they think about, you know, dividing the process and especially the job families and what each family is responsible for.
16:03And so we allow complete customization of kind of what are the tasks, what are the objectives, what are the roles, who gets assigned to those things, and then what do those things actually guide the user through?
16:14And we've got a huge kind of like no-code configuration platform that actually supports that.
16:18And so that no-code configuration platform supports things like versioning and version management, pipeline protection when you roll out new rules, collaboration between people.
16:26So if I've got like three admins and they're all working on a different piece of the process, they can actually work on different drafts of the configuration and merge those things together.
16:34And so we've invested very heavily in making sure that we have easy to use workflow tooling that, you know, doesn't require somebody to know how to code so that every lender can really tune this to, and then continue to iterate on it.
16:44Like one of the beauties that we've seen with our customers is they can set up a process.
16:48They can learn a bunch about where the bottlenecks are, and then they can improve the way that they actually tool their process as opposed to having to like hire around that problem.
16:56And it sounds like the lift to continue to have different iterations of it because it's a no-code environment.
17:02It sounds very light and user-friendly.
17:05That's exactly right.
17:07It's very much a kind of just click, point and click setup.
17:12And the other nice thing about that is it does allow people to bring people with more mortgage expertise into the actual configuration of the operational process, as opposed to today, where you kind of have to find someone who knows how to code and knows the mortgage process.
17:24And those people, as we all know, are pretty rare and pretty expensive.
17:28Pretty rare and pretty expensive.
17:30Exactly.
17:31Well, Mike, thank you so much for taking us through Vesta LOS.
17:34To our audience, for more information about Vesta, click the link below.
Be the first to comment