FCCPC-readyNDPA-aligned

From messy borrower documents to a scoreable credit file in 60 seconds.

CLNA is the AI agent that ingests bank statement PDFs, handwritten market ledgers, WhatsApp receipts, Pidgin voice notes, and IDs — and outputs a clean borrower file your underwriting team or decisioning engine can act on immediately.

  • No credit-decisioning lock-in
  • Mono open-banking integrated · Stitch / Smile / CRC on the roadmap
app.clna.tech / borrower / AR-2026-04827
Source · D1
GTBank · Statement
02/12NIP CR/PAYSTACK SETT185,400
03/12POS/EKE MKT FRIESLAND42,500
05/12CASH DEP CHINAZA SELF48,000
05/12NIP CR/0801****22152,000
05/12NIP CR/0813****44748,000
05/12NIP CR/0907****11854,000
06/12POS/DANGOTE TEXTILE185,000
07/12NIP CR/PAYSTACK SETT212,600
5 layered deposits flagged →
Extracted credit file
Chinaza Okwu
Okwu Fabrics & Asoebi · Onitsha · 4y 3m
BVN match Layering Voice ≠ ledger Ajo wk 56
Avg weekly inflow₦541k
DSCR3.1×
Signal posture6 green · 3 red
generated in 47s · conf 0.91
Eke-day cycle detectedrecurring +38% every 4 days
How it works

Three steps. No model lock-in.

Built specifically for lenders. Prompts engineered for market handwriting and Nigerian-language voice.

01

Ingest anything

PDFs, ledger photos, NEPA bills, WhatsApp screenshots, 60-second voice notes in Pidgin / Yoruba / Hausa / Igbo. Drop in the dashboard, hit our API, or forward by email.

  • bank-stmt.pdf parsed
  • ledger-wk47.jpg parsed
  • voice-igbo.m4a parsed
  • nin+selfie.pdf parsed
02

Reason across modalities

Detect ajo / esusu cycles, market-day spikes, ghost periods, layered deposits, salary-disguising, cross-document inconsistencies, ID tampering. Severity-ranked, never a black-box score.

Cycles detected4 / 4
Eke-market 4-day cycle96%
Weekly ajo to Mrs Umeh99%
Asoebi seasonality84%
Layered deposit cluster91%
03

Output a clean credit file

Structured JSON + signed human-readable PDF, scored confidence per fact. Push to any decisioning engine — Lendsqr Oraculi, Indicina Decide, your in-house webhook — or hand to a credit analyst with the full audit trail attached.

"signal_posture": "positive_with_reservations",
"confidence": 0.91,
"red_flags": ["layered_deposits"],
"avg_weekly_inflow_ngn": 541000
Why CLNA

The Nigerian-context moat

Generic OCR fails on Eke-market handwriting. Generic ASR misses Pidgin code-switching.

Quality vision
Nigerian-handwriting OCR

Reads the Onitsha, Aba, Kano, and Lagos market-ledger handwriting that generic OCR mangles

5 languages
Pidgin + 4-language ASR

Transcribe English, Pidgin, Yoruba, Hausa, and Igbo — including the code-switching of real voice notes.

48h SLA
FCCPC audit vault

Sealed and hash-chained in Cloudflare R2. One-click pack export when the FCCPC demands records inside 48 hours.

<60s
document to credit file
14
document types day-1
5
languages: EN · Pidgin · YO · HA · IG
SHA-256
hash-chained FCCPC audit vault
Pricing

Pay per borrower file. Naira-denominated.

Add-ons: FCCPC audit-pack export · custom extraction schemas · embedded API for fintechs.

Starter
₦300,000/ mo
Cooperatives & small MFIs
  • 500 borrower files / month
  • 5 document types
  • Email support
  • Sandbox + 1 production key
Start trial
Growth
Most lenders
₦1,500,000/ mo
Digital lenders & BNPL
  • 5,000 borrower files / month
  • All 14 document types + voice
  • REST API + webhooks
  • Mono open-banking included · more bureaus coming
  • Priority Slack support
Start trial
Enterprise
₦15M – ₦80M/ yr
Tier-2 banks & multi-country
  • Unlimited files
  • Dedicated CSM + 99.9% SLA
  • Portfolio-tuned extraction — roadmap
  • On-prem / VPC deployment — roadmap
  • NG data residency & region pinning — roadmap
Contact sales

Overage: ₦500–₦800 per file beyond plan limit. NGN via Paystack; USD invoicing available for cross-border lenders.

See CLNA on your own portfolio.

Send us 25 anonymised borrower bundles. We'll process them end-to-end and walk you through what we found — flags, confidences, missed signals — in a 30-minute call. No commitments.