Carmel Limcaoco spent enough time watching loan officers in Manila squint at blurry bank statements to know the process was broken. In much of the world, credit decisions still depend on hand-scanned tax filings, narrative risk memos, and stacks of paper that no algorithm can easily parse. The problem isn't a lack of data; it's that the data arrives as images, PDFs, and handwritten notes that resist automation.
So Limcaoco, who worked on computational audio projects at Apple, teamed up with Rhea Malhotra, a Stanford AI researcher, to build software that treats messy borrower documents as a solvable machine learning problem. Their startup, Kita, announced in August 2026 it has raised $4.5 million in seed funding led by BoxGroup, with backing from Y Combinator, Golden Gate Ventures, BEENEXT, U.S. News & World Report, and Kaya Founders. Angels from Apple and Mercor also participated.
The San Francisco company, which emerged from Y Combinator's Winter 2026 cohort, has built what it calls AI underwriting infrastructure for lenders in the Philippines, Indonesia, Mexico, and the United States, with South Africa listed among its target markets. The pitch is simple: turn borrower documents into credit assessments in under a minute, using vision-language models that can extract underwriting signals from photographs of tax forms or screenshots of bank balances.
Automating the Document Chase
Kita's platform handles two bottlenecks that plague high-volume lenders. An AI Credit Officer chases missing documents from borrowers via WhatsApp, SMS, and email, switching languages as needed. An AI Underwriter then spreads financials, checks policy rules, and drafts decision-ready memos, citing every figure back to its source document. The company says it has processed more than 100,000 borrower files across its live markets, according to its own reporting, though it has not disclosed which lenders account for the bulk of that volume.
One customer, TRBank, Inc., a rural bank in the Philippines, ran 5,100 loan records through Kita's system and reported field-level accuracy above 97 percent, according to Kita's LinkedIn post in mid-2025. The platform integrates with credit bureaus, identity verification services, and payment processors. It can run as standalone modules or as a full loan origination system, depending on how much of the workflow a lender wants to hand over.
The founders have been vocal about why they started with emerging markets. "We turn messy borrower documents into the data layer that powers lending in emerging and undertapped domestic markets," they wrote in their Y Combinator launch post. Limcaoco, a Manila native who is pursuing a master's in computer science at Stanford, has spent much of the past year speaking about the Philippine fintech landscape. Malhotra, who earned Stanford's Firestone Medal for Excellence in Research in 2025, has published seven AI research papers. Kaya Founders, the only Philippine venture capital firm on Kita's cap table, described Malhotra as "one of the most technically credentialed founders" it has backed.
A Crowded Field, With Room to Specialize
Kita is hardly alone in trying to automate credit underwriting. Kaaj, which builds agentic AI for small business credit intelligence, raised $3.8 million from Kindred Ventures late last year. Heron Data automates document intake for high-volume SMB lenders. Lama AI, focused on community banks, closed a Series A in mid-2025. Resistant AI sells fraud detection tools into lending and KYC workflows.
What distinguishes Kita, at least in its pitch, is the focus on cross-border complexity. The platform describes capabilities that span Mexico's SAT and CFDI tax documents, U.S. SBA 7(a) loans, and CDFI underwriting workflows. The company has not disclosed how many employees it has, though LinkedIn showed about 13 people associated with the startup as of August 2026.
Whether the market is big enough to support multiple venture-backed players is an open question. Lending infrastructure has historically been a grind-it-out business, where incumbents defend their turf with integrations and compliance moats. But if Kita's founders are right that most of the world still underwrites loans by hand, there may be room for more than one way to solve the problem. For now, the startup is betting that a combination of multilingual document parsing and fraud detection will be enough to win over lenders tired of drowning in paper.
