Kita

The AI-native stack for modern underwriting — bespoke to every lending use case. Underwrite borrowers anywhere in the world in minutes.

Where Kita ranks

Current edition · updated Oct 20, 2026 · All 5 rankings

About Kita

In emerging markets, open finance is still nascent. Most of the population is traditionally unbanked, banking APIs don’t exist, and a borrower’s financial history lives in documents: e-wallet records, bank statements, utility bills, and more. Because this data is unstructured, credit and risk teams are forced into manual review. This slows decisioning, increases costs, and caps lending volume.

Kita is the AI platform for global lending operations. We help lenders in emerging markets automate application completion, document verification, and underwriting from messy financial documents — using AI to extract fraud-checked data and localized risk signals that power faster, better credit decisions.

Under the hood, Kita is a learning engine. We link document-level signals to repayment outcomes, allowing our models to continuously improve fraud detection and risk assessment over time. This creates a compounding advantage for lenders as their distinct underwriting decisions feed back into the system.

We’re Carmel and Rhea. We met before Stanford and have been building together ever since. Carmel is from Manila, is a repeat founder, and spent three years in product at Apple. Rhea has a research background in computer vision and received the highest honor in Stanford Computer Science. Together, we combine deep local context with strong technical execution to build the infrastructure that expands access to credit in emerging markets.

Products

What customers say

“TRBank’s lending archive was, in practical terms, unqueryable: thousands of credit folders accumulated over years, each running to several hundred pages of typed forms, handwritten notes, collateral appraisals and bureau reports.Working with Kita, a specialist document AI platform built for lenders, we processed 5,100 credit folders spanning 2022 to 2025, structuring over 450,000 data points with independently verified accuracy above 97%. PHP 8.77 billion of approved loan value is now fully queryable for the first time: the bank’s own ground-truth record of what it lends, to whom, on what terms and against what collateral.”
CustomerKita customer
  • Kita Capture

    N90 uses Kita Capture through its API to extract lender-specified data from Philippine borrower documents. Kita reports more than 43,000 pages processed since February 2026, 98.5% document processing success, and reduced…

    Case study
  • Kita Capture

    TRBank used Kita Capture to structure its historical paper credit folders for search and analysis. Kita reports processing 5,100 loan records into 452,236 structured field values in less than a week, with 97%+…

    Case study

Security and compliance

As reported by Kita. Ask for audit reports before you buy.

  • ISO/IEC 27001In progress
  • SOC 2 Type IIIn progress
  • Trust Services CriteriaCompliant
  • AES-256 encryption at rest
  • TLS 1.3 encryption in transit
  • Automatic key rotation in managed KMS with split-knowledge access
  • SSO via SAML 2.0 (Okta, Azure AD, Google Workspace) on Enterprise
  • MFA enforced for all human access
  • Role-based permissions
  • Scoped API keys per environment
  • Session timeouts
  • Private VPC networking, no public ingress to compute
  • Edge WAF and DDoS protection
  • Egress allowlists for outbound traffic
  • Strict logical tenant isolation; no data, compute, or model context shared across tenants
  • Dedicated environments available on Enterprise
  • Audit logging with timestamps, user, and request ID; tamper-evident; retained 12+ months
  • Logs streamable to customer SIEM
  • Threat-modeling on every new surface
  • Mandatory peer review and security review for auth, encryption, or data egress changes
  • Branch protection on every repository
  • Automated SAST on every pull request
  • Dependency vulnerability scanning blocking known CVEs
  • Container image scanning at build and runtime
  • No long-lived credentials in code; all secrets vaulted via short-lived tokens
  • Secrets rotated on fixed schedule and on personnel changes
  • Locked dependency manifests, signed builds, reproducible CI
  • SBOM available on request
  • Third-party libraries reviewed before introduction
  • Pre-signed URLs scoped to single upload, never reusable
  • Single-tenant isolated compute for document processing; no cross-tenant context or shared memory
  • Encrypted backups with point-in-time recovery
  • Restoration tested on documented schedule
  • Multi-AZ architecture by default
  • Regional failover playbooks exercised regularly
  • Documented incident response plan with severity levels, on-call rotation, and escalation paths
  • Customer notification typically within 72 hours of confirmed impact
  • Annual third-party penetration tests
  • Continuous internal scanning
  • Responsible disclosure program
  • Background-checked engineers with production access
  • Annual security and privacy training for all team members
  • Secure-coding training for engineers
  • NDAs and confidentiality clauses before access to customer data
  • Least-privilege, time-bound, MFA-gated production access
  • Access revoked same day on offboarding; quarterly access reviews
  • Customer-managed encryption keys (BYOK) scoped per deployment
  • VPC and on-premises deployments available on Enterprise
  • SCIM-based provisioning available on request

Is this your company?

Claim the profile to keep facts current, answer buyers and submit evidence for the next ranking review.

Claim Kita