Use case
Kederi for lenders and loan teams
A loan team reads an applicant's statement to answer a few questions. How much comes in, and from whom? How much of it is the applicant's own money moving between accounts? Does the applicant repay other loans, or spend on betting?
On a Nigerian statement, most of those answers are in free text that a person types into a transfer. Kederi reads each row and returns the party, the purpose and a category. The people, accounts and references in the examples are invented. The businesses are real merchants in our directory. A test runs each one through the engine and compares each answer on this page.
What comes back for each row
Send each narration with its direction, credit or debit. The direction comes from the statement, and some narrations name a different party on a credit.
A salary credit
NIP/ZENITH/BRIGHTWAVE LOGISTICS LTD/OCTOBER SALARYName Brightwave Logistics Ltd, intent Salary. The name is the employer, so a team can see that the salary comes from one payer each month.
A repayment to a digital lender
FAIRMONEY LOAN REPAYMENTName FairMoney, category Loan.
A loan paid to a person
NIP TRANSFER TO CHIDI NWOSU LOAN REPAYMENTCounterparty Chidi Nwosu, intent Loan. The purpose comes from the words that the sender typed, and it is not part of the name.
Betting
SPORTYBET NIGERIA LIMITEDName SportyBet, category Betting, MCC 7995.
What comes back for the whole statement
Send all the rows of one account to POST /reports. These reports answer the questions above:
counterparties looks for the account holder in the narrations where the holder is the sender. If it finds the holder, it puts the money that moved under the holder's own name in self_transfers, apart from income. If it does not, account_holder is null, and no money is counted as a self transfer. It also reports money that came through a payment service whose narration does not name the sender, as unidentified_inflow.
summary gives the totals in and out, and the money out for each category, for example Loan and Betting.
subscriptions lists the payments that repeat, with the amount per month and the next date.
forecast gives a range for the money in and out over the next 30 days, or over a period that you choose. The range has a low, a middle and a high figure.
What Kederi does not do
Kederi gives no credit score and makes no loan decision. It reads the rows, and your team or your model decides. If a narration names nobody, the answer says so. If Kederi is not sure of a name, the answer has low: true and a low confidence, so your rules can send that row to a person.
Kederi stores no transaction. The applicant's rows are in your request, and the answer comes back. Nothing stays on our side to read later. A correction that you send to POST /feedback is the one exception, and only when you send it. See the privacy page.
Try it on a statement
The playground reads one narration with no account. The API reference has each request in six languages. To talk about your volume, contact us.
Sign up