Kita

Kita is a document intelligence platform for lenders and financial institutions in emerging markets that converts messy borrower documents into fraud-checked, decision-ready risk signals for underwriting.

Key capabilities include:

• Document capture and verification workflows via a Credit Agent that manages document back-and-forth during loan origination, including WhatsApp and email follow-ups to reduce incomplete submissions and borrower drop-off
• Vision-language model extraction with Capture that reads low-quality inputs (photos, scans, screenshots) and outputs structured data in JSON, CSV, or Excel
• Document-level fraud detection and risk assessment calibrated to local formats across bank statements, e-wallet records, government IDs, tax documents, utility bills, payslips, mobile money records, and insurance documents
• Integrations for loan origination systems (LOS) through a Web UI (including a white-label portal) and an API with published documentation

Pricing: Usage-based pricing model based on document volume, featuring a free tier. For Growth and Enterprise plans, contact for pricing.

Notable for its hyperlocalized focus on emerging-market document formats (for example Southeast Asia and Latin America) and for using computer vision and vision-language models to handle noisy, non-standard documents that traditional OCR often fails to parse. Backed by Y Combinator (W26).