Mi Casita - Financial Tool for Latino Immigrants in the US

Inspiration

For millions of Latino immigrant families, money is never just personal. It is a paycheck in New York that pays rent here and groceries for a mother in Guatemala, Colombia, or Mexico. Yet the tools built to help people manage money assume you have a credit history, a Social Security number, a bank account, and perfect English.

We kept coming back to one question. Where does the money quietly disappear? The answer was everywhere. It disappears in exchange rate markups that nobody sees, in fake "IRS" letters demanding gift cards, in "notarios" charging thousands for immigration help they can't legally give, and in payday loans taken out the week rent comes due.

We wanted to build something that feels like a casita, a small, safe home for your money. It had to be simple, trustworthy, and in your language.

What it does

Mi Casita is a voice-first financial tool for Latino immigrants in the US. It never asks for an SSN, ITIN, or bank account.

True cost of sending money home. Users enter what they send and which provider they use. Mi Casita reveals the hidden exchange rate markup and shows, with math instead of advice, what switching providers would save over a year.
Voice and text ledger. A small business owner can say "hoy vendí 40 tacos a 3 dólares y gasté 60 en carne" and the app logs it and builds a monthly profit and loss statement. Those records later help with taxes and small business loans.
Letter scanner. Snap a photo of an English letter from the IRS, a bank, a landlord, or a debt collector. Mi Casita explains it in simple Spanish, pulls out the deadline and next steps, and flags scam warning signs.
Quincena, the paycheck forecast. Hourly workers enter their shifts and bills, and the app predicts which week they will run short. It then shows safer options than a payday loan, like credit union alternative loans and lending circles (tandas).
Learn. Short financial literacy lessons with a read aloud button, for users more comfortable listening than reading. (Not Financial Advice).

Everything is educational. We show the numbers and the options, and the user decides.

How we built it

We started by prototyping the full experience as a standalone HTML demo to lock in the flows, the math, and the visual identity before spending any build credits. Then we rebuilt it in Base44 within 4 hours on a 125 credit budget, using its AI integrations for document reading, speech input, and text to speech.

The remittance model is deliberately simple so anyone can verify it. For a transfer of amount $A$ with a flat fee $F$ and a hidden exchange rate markup $m$, the real cost is

$$ \text{True cost} = F + A \cdot m $$

and the family receives

$$ \text{Received} = A \cdot r \cdot (1 - m) $$

where $r$ is the real mid-market exchange rate. Sending \$200 twice a month with a \$5 fee and a 3.5% markup costs \$12 per transfer, or

$$ 12 \times 24 = \$288 \text{ per year} $$

and more than half of that is invisible to the sender.

For the paycheck forecast, we simulate the user's balance day by day over the next 4 weeks using paydays, bills, daily spending, and transfers home, then flag the first week where the balance drops below zero. To put payday loans in perspective, a typical fee of \$15 per \$100 borrowed every 2 weeks works out to

$$ \text{APR} \approx 0.15 \times 26 = 390\% $$

Our color system was designed for trust. Deep navy signals stability, green marks money that is yours, warm gold highlights hidden costs, and red appears only for real danger like scams or shortfalls.

Challenges we ran into

Regulation. We can't give financial advice, and credit building and wage advances are heavily regulated. We designed every feature to show math and options instead of recommendations.
A tight credit budget. With 125 credits, every prompt mattered. Prototyping outside Base44 first let us write detailed prompts that landed close to right on the first try.
Designing for trust. Early testers found our named assistant character distracting, so we replaced it with a neutral design. The tool should feel like a trusted home, not a persona.
Real speech is messy. People say "dos cincuenta" to mean \$2.50, not 52. Fine tuning the languages took more care than we expected. Many immigrants from Guatemala and southern Mexico speak Aztecan languages like Nahuatel first, and some don't read Spanish comfortably. That pushed us toward voice input, a text box as backup, and read aloud everywhere. This is an area for development.

Accomplishments that we're proud of

Turning an invisible cost into a single number people can see and act on.
A letter scanner that can tell the difference between a real IRS notice and a gift card scam.
A trust-first design that asks for nothing sensitive.
Building a working, multi-feature product in 4 hours.

What we learned

We learned that for underserved communities, trust and simplicity matter more than features. A tool that asks for an SSN on the first screen loses the people who need it most. We also learned how much of financial harm comes from information gaps rather than bad decisions, like a fee hidden in an exchange rate or a letter written in confusing English. And we learned to respect the legal line between education and advice, which shaped every design choice.

What's next for Mi Casita - Financial Tool for Latino Immigrants in the US

Indigenous language support for Nahuatel, and other Aztecan languages, built in partnership with community organizations.
Live remittance quotes so comparisons use real time fees and rates.
Narrated video lessons for the Learn section.
Exportable records so a small business owner's ledger can be shared with a tax preparer or lender.
Partnerships with credit unions, lending circle nonprofits, and free tax clinics, so the options we show become one tap away.