Kita bets AI can read messy loan documents; $4.5M seed, $130M processed

Kita is an AI credit platform for banks, fintechs and microfinance lenders that turns borrower documents into decision-ready credit memos.

In markets with no banking APIs, borrower histories live in messy documents, so AI that reads any format can become lending's data layer — fast and fraud-checked.

$4.5M seed led by BoxGroup (announced August 2026), with Y Combinator, Golden Gate Ventures, BEENEXT, Kaya Founders and Apex Star Capital participating

Carmel Limcaoco (from Manila, ex-Apple) and Rhea Malhotra (ex-Stanford computer vision, deferred a Princeton AI doctorate) met as Stanford classmates during COVID and founded Kita in 2025, after seeing Manila lenders buried in loan paperwork.

Within five months Kita said it processed over $130 million in loan volume, more than 60% of it from the Philippines; for rural lender TRBank it converted four years of paper lending history into searchable data in under three days at above 97% verified accuracy. It joined Y Combinator's W26 batch — the company says Limcaoco is the first Filipina founder admitted in over five years — and in August 2026 closed a $4.5M seed led by BoxGroup.

Still live: Kita is processing loans across the Philippines, Indonesia, Mexico and the US, and is using the seed to expand its engineering team and underwriting and fraud-detection tools as it pursues customers across Southeast Asia, Latin America, Africa and the US.

Philippine lenders face two problems at once: badly backlogged manual review and too little data to decide confidently — the two things document AI directly attacks.

Emerging markets lack banking APIs, so a borrower's financial history is locked in noisy documents; whoever can read them well owns underwriting.

Going after the messiest archives first — handwritten passbooks, crumpled payslips, e-wallet screenshots — built a moat that clean-data OCR tools cannot cross.

Founder credibility (Stanford CS, Apple, YC) plus investors like BoxGroup, Golden Gate Ventures and Apex Star Capital financed the long sell to conservative lenders.

Keeping humans in the loop on final decisions removed the regulatory and trust objections that block pure-automation credit products.

When incumbents can't read their own data, the wedge is reading what others skip: start with the messiest archives, not the cleanest APIs, and make every output traceable.

As of late August 2026, Kita was live with lenders across the Philippines, Indonesia, Mexico and the US, using the $4.5M seed to expand engineering and underwriting and fraud tooling while hiring in the Philippines. It faces competition from Kaya Founders-backed Moneta and other document-AI credit startups, but argues its no-configuration reading of unfamiliar formats and source-linked outputs differentiate it. No revenue or valuation figures were disclosed.