Odysı
See if we fit
Solutions
Solution Lending & mortgage

AI for mortgage and lending qualification

Pre-qualification is repetitive, rules-based work: applying the same criteria consistently to decide who can realistically proceed. That makes it worth automating, as long as the lending decision itself stays with a person.

O
Odysi Solutions
Proof Urban Capital, in production
Read 6 min

Mortgage and lending advisory starts with a screening question: can this person actually qualify, and for how much? Answering it well takes a set of criteria applied consistently, income, debts, savings, employment, and, in markets like Spain, the region, since regional taxes and fees change how much cash a buyer really needs. This is a practical look at where automation fits, what it should not decide, and how we approached it for a mortgage advisory firm in Spain.

01

Where advisor time goes before automation

For a small advisory team, the expensive part is screening. An advisor judges from income, debts, savings, employment, and region whether a prospect could qualify, and every hour spent on someone who cannot is an hour away from someone who can. When leads live in a spreadsheet with no scoring and no record of why someone was or was not a fit, that judgment has to be repeated and cannot be reviewed later.

There is a second cost that is easy to miss. The real cash a buyer needs sits well above the headline down payment once regional taxes and fees are included, so prospects are often caught out late in the process. Making that number clear early changes the quality of every conversation that follows.

Fig. 1 · The real cost Example · a home purchase
What buyers expectdown payment
What they actually need+ regional taxes & fees
The teal portion is the gap: the extra cash a buyer needs beyond the down payment. Showing it early prevents late surprises.
Fig. 1: the real cash needed sits above the headline down payment once regional taxes and fees are counted.
02

What automation can do, and what it should not

Pre-qualification is a strong fit for a self-serve, rules-based system. It can let a prospect self-assess against the same criteria an advisor would apply, calculate the real cash required including regional taxes and fees, produce a scored lead so the team sees who is worth a call and why, and keep a record of the inputs and reasoning behind each result.

It should not make the lending decision. Qualifying a mortgage is a regulated activity, and a self-assessment is a filter and a first estimate, not an approval. The system screens and informs so a human advisor spends time on prospects who can realistically proceed. The decision, and the regulated advice, stay with a person.

In a regulated field, being able to explain why a result came out the way it did matters more than sophistication. Transparent scoring is a feature, not a limitation.

03

What a sound system looks like

self-assessment scored lead advisor
The parts that matter The qualification criteria as the source of truth A mobile-first self-assessment A transparent cost breakdown, shown early Lead scoring that records its reasoning A human advisor on the decision
Fig. 2: the system screens and scores; the advisor gives the advice and makes the decision.
04 · In production

What we built for Urban Capital

Urban Capital is a mortgage advisory firm in Spain. We designed, built, and run a mobile-first web app that lets prospects self-assess their eligibility, see the real cash they will need, and arrive as a scored lead, all before an advisor picks up the phone. It replaced hand-screening every prospect and a spreadsheet with no scoring. It is live in production.

Read the Urban Capital case study
05

How to start without overcommitting

Begin with the screening you already do by hand, and encode those exact criteria rather than inventing new ones. Show prospects the real cost early, score the leads, and keep the advisor's judgment on the actual decision. The aim is narrow: stop spending advisor hours on prospects who cannot qualify, and give the ones who can a clearer, faster start.

Common questions

FAQ: AI in mortgage qualification

Can AI approve a mortgage or a loan?
No. Qualifying and approving a mortgage is a regulated decision made by a lender and a qualified advisor. A pre-qualification system screens and estimates so the right prospects reach a person; it does not approve anything.
What can automation handle in mortgage qualification?
Self-assessment against the firm's criteria, calculating the real cash a buyer needs including regional taxes and fees, and scoring leads so the team knows who to call and why.
Does it replace mortgage advisors?
No. It removes the manual screening that consumes advisor time, so advisors focus on prospects who can realistically proceed and on the advice itself.
Is transparent scoring better than a complex model here?
In a regulated field, usually yes. Being able to explain why a prospect scored the way they did is more valuable than an opaque prediction.
Does it work outside Spain?
The approach applies to any market. The specifics, such as which taxes and fees apply and what the qualification criteria are, are configured per market.
Keep reading
Free field assessment Is your AI project worth building? Take the AI Project Scorecard. Five questions, two minutes, an honest verdict: build, fix, or wait. Take the test →
Prototype. Automate. Grow.

Screening every prospect by hand?

If your team screens every prospect by hand and leads sit in a spreadsheet with no scoring, that screening is the part worth automating, while advice and the decision stay with your advisors. We are easy to talk to.