Couverture de The AI Edge: Transforming Non-QM Lending

The AI Edge: Transforming Non-QM Lending

The AI Edge: Transforming Non-QM Lending

Écouter gratuitement

Voir les détails
HOW DO YOU AUTOMATE UNDERWRITING WHEN THERE IS NO RULEBOOKHere is the kind of question Sarah Gonzalez was willing to hand to software. In the example she uses, a matrix calls for an 800 FICO on a million-dollar loan and caps the loan-to-value at 75, and the file in front of the underwriter arrives at 76. That one has an answer before anyone reads the file. Almost nothing else in her book does, because almost everything Logan Finance writes is non-QM, and there is no standard guideline set sitting there waiting to be encoded.Gonzalez is Chief Innovation Officer at Logan, and she has had to answer the automation question in production rather than on a panel.CAN YOU AUTOMATE UNDERWRITING WITHOUT AUTOMATING THE CREDIT DECISIONLogan built a tool called Aura. A partner supplies the baseline rules engine, and the rules themselves were written in-house and belong to Logan. That distinction is worth holding onto, because renting an engine and owning the rules inside it is a different asset from renting the whole thing.What Aura does is narrow on purpose. It reads hundreds of pages of guidelines so that an underwriter does not have to page through matrix after matrix, it reports what sits inside the box and what sits outside it, and then it stops. It does not approve anything and it does not decline anything. Gonzalez is strict about that boundary in a way most vendors selling into this market are not.If the tool is already reading the guidelines, does the judgment itself move to the machine within a year or two? WHY SHE DOES NOT START WITH THE ORG CHARTHer panel at the HousingWire AI Summit was about the future org chart, and her answer inside her own organization is that she does not start there. She starts with culture and with the experience she wants the customer or the employee to have, then asks which parts of that work can be automated and which parts require judgment.WHAT SHE WANTS FROM VENDORS, AND IS NOT GETTINGThe sharpest stretch of the conversation is not about her own build at all. It is about everyone selling to her.She arrived at the recording from a week of vendor demos. Her question at the end of each one was not about the model, it was about the implementation timeline and the implementation cost, and the answers kept coming back at six months and a price she thought the work no longer justified.WHAT YOU'LL TAKE AWAY- The automatable surface in non-QM is eligibility, not decisioning, and the two are worth separating in writing before anyone builds anything- A rules engine you rent with rules you own is a different asset from a platform you rent entirely- The honest return on an eligibility engine is underwriter throughput, and hers is 20 to 25%- Eliminating a task and eliminating a person are two separate decisions, and conflating them is a leadership failure rather than a technology one- If a vendor cannot tell you why implementation takes six months, that is the question to keep askingCHAPTERS(00:00) What AI can't do in non-QM lending(01:35) An operator handed the innovation job(03:29) Taking a technology role without a technology background(06:25) Automating a lender with no rulebook(09:22) Will judgment ever move to the machine(11:29) Why she doesn't start with the org chart(12:24) Eliminating task, not judgment(14:01) What the manager's job becomes(16:27) Why vendor AI still takes six months(21:18) Inside Aura, eligibility not decisioning(22:48) A 20 to 25% lift for underwriters(25:07) The right partner at the right costMENTIONED IN THIS EPISODELogan Finance, the non-QM lender where Sarah Gonzalez is Chief Innovation OfficerAura, Logan's eligibility engine, built on a partner's rules engine with rules Logan wrote and ownsHousingWire AI Summit, the Dallas event in August 2026 where the hosts met herThe Mortgage Collaborative, whose summit she had attended the week before recordingIDP and OCR, the document technologies she argues get folded into the word AIClaude, named as an example of what per-seat AI costs a mid-size lender without change management around itRE/MAX, Redfin and Rocket, named by Reuven as evidence the ground is shifting upstream and downstreamABOUT THE GUESTSarah Gonzalez is Chief Innovation Officer at Logan Finance. She has spent thirty years in mortgage, including chief operating and president roles at earlier companies, and now works on where technology genuinely changes how lending gets done rather than where it only adds a step.LinkedIn: https://www.linkedin.com/in/sarahbatangan78/YOUR HOSTSChris Grimes is CEO of FundMore.Reuven Gorsht is CEO of Deeded and The Variable.One builds the tool. One absorbs the friction.LISTEN AND SUBSCRIBECaptivate — https://the-signal.captivate.fm/Apple Podcasts — https://podcasts.apple.com/us/podcast/the-signal/id6805026259Spotify — https://open.spotify.com/show/59UmJswLjpRO5p7vUwr7hVYouTube — https://www.youtube.com/@thesignalreuvenchrisNew episodes every Tuesday.
adbl_web_anon_alc_button_suppression_t1
Aucun commentaire pour le moment