Skip to content
Consonant

Kasho

In development

Financial intelligence infrastructure for African banks and fintechs.

Built by the studio.

Talk to us about Kasho
from raw descriptor to structured merchant
< 200 ms
from raw descriptor to structured merchant
Nigerian rails covered: POS, NIP, USSD, ATM, card
5
Nigerian rails covered: POS, NIP, USSD, ATM, card

The most-viewed screen in every Nigerian banking app is unreadable: terminal IDs, NIP session strings, USSD codes, truncated POS descriptors. Until that data is classified, nothing intelligent can be built on it, whether that is spending insights, subscription detection, credit signals or any AI feature on a bank’s roadmap. Kasho is the layer that makes it readable, then makes it useful. It is two products on one data engine: Enrich and Advisory.

What it does

  • Enrich: a raw descriptor in, and the merchant name, logo, category, branch, coordinates and a subscription flag out, in under 200 ms
  • Every Nigerian rail: POS, NIP, USSD, ATM and card-not-present
  • Three layers of classification: known-merchant patterns, machine learning, and Kasho Scout, a field network that verifies the businesses no registry lists
  • Advisory: an AI layer inside the institution’s own app that profiles each customer’s income cycle, spending, idle balance and savings rate
  • Product recommendations from the institution’s own catalogue, with amounts worked out from the customer’s real data
  • The institution keeps execution and authentication. Kasho never holds credentials

REST API · < 200 ms · NDPA-compliant · deployable inside the institution’s perimeter · LLM tool-calling

The prerequisite, as an API
Budgets, insights, subscription detection, credit risk and every AI feature depend on classified transactions. Banks should not have to build a Nigerian merchant classifier to get them. Enrich is that classifier, one API call per transaction.
A merchant directory nobody else has
POS terminals in Nigeria now report GPS coordinates, so Kasho knows where every unidentified merchant is. Scout agents visit, photograph and record the business, and from then on every transaction at that terminal resolves instantly, for every partner.
One reasoning engine, any catalogue
Advisory runs on tool-calling. One AI engine reasons about the customer; each institution’s products, rates and eligibility rules are configured as tools. Adding a bank means configuring tools, not training a model.
It compounds
Every partner and every transaction sharpens classification for all of them, and every month of Advisory makes its guidance more personal. That is the switching cost fast transfers never gave banks.

The most-viewed screen in banking, made readable.

Kasho Enrich turns a raw Nigerian transaction string into a merchant, category, location and subscription flag in under 200 milliseconds. Kasho Advisory builds on it.

Descriptor

POS PURCHASE SHOPRITE ILUPEJU LA NG

POST /enrich

{
  "merchant": {
    "name": "Shoprite",
    "branch": "Ilupeju",
    "category": "groceries",
    "coordinates": { "lat": 6.5528, "lng": 3.3515 }
  },
  "classification": {
    "category": "groceries",
    "subcategory": "supermarket",
    "confidence": 0.97
  },
  "subscription": {
    "is_subscription": false,
    "is_recurring": false
  }
}
Illustrative request and response.

More from the studio