The documents are photos
Customers send ID cards shot on a kitchen table, statements printed and rescanned, proofs of address folded in quarters. Glare and skew are the norm, not the exception.
IDs, proofs of address, payslips and bank statements come in as photos and scans. Arakma reads them, reconciles the Arabic and Latin spellings of a name, checks the documents against each other, and hands your analyst only the cases that need judgement.
For banks, microfinance institutions, fintechs, leasing and consumer-credit teams.
Where the time goes today
Customers send ID cards shot on a kitchen table, statements printed and rescanned, proofs of address folded in quarters. Glare and skew are the norm, not the exception.
يوسف on the ID, Youssef on the payslip, Yousef on the bank statement. Matching them by hand is slow, and getting it wrong is a compliance finding.
The people trained to judge risk are copying income figures from statements into a spreadsheet. Applications wait, and customers go to whoever answers first.
How Arakma handles it
The same four steps for every document, set up with your fields, your checks and your thresholds during a pilot on your own files.
From your app, a branch scanner or an email. Front and back of the ID, statements of any length, in any combination.
Names in Arabic and Latin script, ID number, dates and the MRZ on passports. The two spellings of a name are reconciled rather than treated as two people.
Name and address agree across the file, the ID is in date, the proof of address is recent, and income on the statement is compared with what was declared.
Anything uncertain or inconsistent is held with the reason shown. The complete, checked file goes to your onboarding or credit system.
What goes in
Scanned, photographed or born digital, in Arabic, French or English. If yours is not listed, it is not a special case: describe the fields and we read it.
Where it goes
Excel, CSV or JSON, laid out to match your import template. A direct connection is built when your system allows one.
What changes
A complete file is ready for a decision as soon as it is uploaded, instead of waiting in an analyst's queue.
Every extracted value, every check and every human correction is recorded with who made it and when, and can be exported for an inspection.
Documents are stored encrypted in the EU, and never used to train a model.
Questions
Anything else, including security questionnaires, we answer before a pilot. How we handle data.
No. It reads and checks the documents and tells you what it found and how sure it is. The decision stays with your analyst and your policy.
Yes. Arabic and Latin script are read natively, the MRZ is parsed and checked, and the two forms of a name are linked in the output.
In the region you choose when the account is created, encrypted, and deleted after your retention period, 30 days by default.
We run them through Arakma with your fields and show you the output, scores and review decisions in a call. If it works on yours, we scope a pilot.
Or write to contact@arakma.com. We reply within one business day.