Uptiq

Transcription:

Kyle Kneubuhl (00:12):

Hello, everyone. I'm Kyle Kneubuhl, and this is my colleague, James Hallacy. We're from Uptiq, the AI platform for financial services.

Full disclosure, I'm not a tech guy. I'm a commercial lender of over 20 years. It took me a week to learn how to spell "LLM." Today, we're going to walk you through one of our agents that handles the entire commercial loan process. James, take it away.

James Hallacy (00:49):

Absolutely. I'm logged into our system as a relationship manager at a bank, and I'm going to use one of our preset prompts to get the flow started. I just got off the phone with my client and want to start a new loan application that will kick off the agent.

In this case, I've collected a loan application. You can see a standard application here, and I'm going to upload it. This is ultimately going to kick off the autonomous agents. This is where a lot of the underlying action is happening right now. It is extracting data from this application, pulling out the guarantor structure, and understanding the context of the loan. For example, because it's a commercial real estate loan, what documents do I need to collect? What analysis will I need to do?

Ultimately, it looks at the credit policy and puts together the details to make decisions and move the loan forward. In this case, we used a loan application form to extract that information. However, that could be any form of intake, whether it's coming from an LOS or another system.

Right now, it's going to move me over to the app details, where it's created the customer profile and sent them their login information. You can see here that we've pulled out the ownership structure. I can look at my entities and, if I need to, I can edit them. I'm not going to do that today, but I am going to move over to our document agent. Kyle, do you want to talk more about the document agent?

Kyle Kneubuhl (02:38):

Absolutely. Because the agent understands the bank's credit and product guidelines, it also establishes what documents are needed to complete a package for underwriting. Under each of the entities—borrower and guarantor—the documents are highlighted. From here, it can either automatically and autonomously gather those documents from the borrowers and guarantors or be prompted by the banker. Or, if you have clients who don't want to deal with tech and just say, "Here are all my documents," we can do a bulk drop within the document queue, as James is going to do now.

James Hallacy (03:12):

So let's say I have my document package. I, the borrower, have uploaded my data room. It's actually going to use that agent to appropriately identify if I have everything right to move forward. For example, here we have the Collier balance sheet. If I take a look, what is our agent going to do? It's going to look and make sure this is a balance sheet. It's going to see that this is for Oak Grove Professional, and it's going to identify that it was supposed to be for Oak Grove, our other guarantor. So ultimately, it's going to verify if I have everything I need and completely reduce the back-and-forth that the loan officer has to do to get this loan to a point of funding and analysis.

Once I confirm those documents, it begins the extraction process to move over to underwriting.

Kyle Kneubuhl (04:08):

Correct.

James Hallacy (04:08):

I'm actually going to jump over here into the credit analyst view, where you can see some of the underwriting.

Kyle Kneubuhl (04:15):

Perfect. It's important to note that once those documents are in, it will also connect directly with your repository or LOS system and push and pull the documents so we aren't doing duplicate tasks. From here, the documents have been extracted, along with key data points from the credit guidelines and industry standards for the loan deal, and it's actually spread them out.

As you can see, this might look familiar to a lot of people. This is a real case from one of our clients who wanted to use their Excel workbook for these manipulations. They're familiar with it, so there's no training. It's designed to adapt the way you want to see it, so that's their user interface. From there, each of the spread cells can be changed. Comments can be added about why it was changed and what the mitigation was. At any time, if you want to find out where the spreading information came from, you can right-click and go to the source documents instead of parsing through pages.

At any time, if you want to know what the agent did autonomously on your behalf, there's a "journey" button that will tell you exactly what the AI did, why it did it, and where it pulled that information from in your credit policy to give you the correct output.

From here, any manipulation, comments, or changes flow through the credit tables and generate a credit memorandum all in one place. Gone are the days of multiple pages where you have to make sure you changed something here and it was updated there, avoiding errors. It will flow through and the CM will use the same template as the institution and be filled out as if one of their employees did it on their behalf. That also takes into account notes, comments, and generative AI summaries of findings within the tables.

Obviously, that will get you to a decision and funding much faster, but let's talk about what happens after funding.

James Hallacy (06:06):

After we've funded the loan, a lot of servicing goes into this, including covenant tracking and collecting documentation. We have created the ability to upload your loan documents. I'm going to show you an example of a basic Laser Pro loan document. I'm sure you've all seen a hundred of these before. We upload that, and we use our fine-tuned models to begin to extract the information. Our models have appropriately identified all the covenants based on the document—financial and negative—for each of our guarantors. They've given us the type of information we'll need, when we'll need it, where it is, the status, a brief synopsis, and whether or not it's been reviewed. That works for the financials as well as any of the negative information.

Kyle Kneubuhl (07:04):

Yeah, so this live reminder system will know before, during, and after reporting deadlines and will interact with the client to get it more timely. It will also audit those financials to make sure they're correct and that your list is compliant. Once the documents are in, it will extract and spread them to ensure compliance while giving you insights if anything is off track, allowing you to be proactive instead of reactive on any troubled credits.

Basically, it's giving you a live health understanding of your compliance and portfolio without having to scale by adding more people to monitor it and wasting your SMEs' time on tracking documents. It does it all in real-time.

With that, we've walked you through one of the most complex workflows within any financial institution. We understand that every institution defines AI differently depending on their focus and operations. That's why our platform doesn't force you to change the way you work. Instead, it adapts to your process, enhances it, eliminates redundancies, and fits seamlessly within the systems and workflows you use today.

So when you think about Uptiq, who we are, what we do, and what sets us apart, just remember what we consistently deliver to our clients: Real AI, real expertise, real results. Thank you.