- Home
- For Banks
Industry solution · banking
Data control stays with the bank: the entire stack runs inside your perimeter
Services, database, files, search, and AI deploy on the bank's servers and never leave the perimeter. The flagship is the records archive: legacy records enter the system in batches, scans become available to full-text search, and destruction happens only under an approved act.
Typical scenario · bank archive
Legacy records enter the system in batches and become searchable
The bank's paper and digital legacy - case files in boxes, scans in network folders, registers in spreadsheets - follows a single managed route: from batch intake to full-text search and scheduled destruction.
step 1
Legacy
Paper case files, scans in network folders, and registers in spreadsheets are exported in batches: inventories, case files, volumes, files, and links.
step 2
Dry run
The batch is validated in full before any write: the entire reference graph is checked, and a single verdict covers the whole batch.
step 3
Discrepancy report
Discrepancies are visible before anything is written; the archivist decides what to fix at the source. Re-uploading a batch does not create duplicates.
step 4
OCR
OCR: 8–12 thousand pages per hour on four cores; the error rate for Russian and Uzbek Latin script is 0.2–0.3 %. We measure on your samples during the first week of deployment.
step 5
Permission-aware search
Recognized text is indexed the same way as cards and files: an employee finds only what their classification level and clearance allow.
step 6
Appraisal and destruction
When retention periods expire, records appraisal follows; a volume is destroyed only against its own line in an approved destruction act.
The storage lifecycle - the records classification scheme, volumes and case files, acts - is covered in detail on the Records Archive page.
Security and compliance
Compliance gets a mechanism, not a promise
Every item below is a working product mechanism: it can be demonstrated in a live environment and recorded in a security questionnaire.
- SHA-256 for every file version. The checksum is computed on every change and recorded in an immutable audit log: file substitution shows up on verification, and the action history cannot be rewritten after the fact.
- Classification levels 0–4 and personal clearances. Five confidentiality levels: a document opens only for an employee whose clearance matches its classification level - a stray link grants nothing.
- ACL: "Deny overrides Allow." An explicit deny is stronger than any allow - an accidentally granted role does not open a document closed by a direct deny.
- User directory - LDAP / Active Directory. Accounts, attributes, and lockouts sync from your domain. A departed employee loses access to KASKAD the moment they are disabled in the domain - no manual second registry.
- Personal data localization. Documents, personal data of employees and clients, and logs physically reside on your servers in Uzbekistan. The on-premise architecture satisfies the localization requirement (Law of the Republic of Uzbekistan "On Personal Data", ZRU-547) by design: there is nowhere for the data to leak.
The full model - roles, backups, updates without a weekend maintenance window - is on the Security page.
Head office and branches
One installation for the entire branch network - the database enforces isolation
A multi-tenant architecture put in banking terms: branches work in a single system, while unit data is separated by the DBMS, not by a condition in code.
One installation
The head organization and branches share a single system: unified reference data, roles, routes, and one audit log instead of a zoo of local copies.
Isolation at the DBMS level
Row-level security in PostgreSQL: the database itself isolates each unit's data. A query without the right permissions returns no foreign rows, by construction.
Flexible licensing
A modular license bound to the customer's server: you enable the circuits you need, and workstations are counted per user. Branches run on a single installation - no separate deployment in each.
See the bank archive in a live environment
The demo takes 40 minutes and is tailored to your case: legacy migration, OCR, search, retention rules. Tell us about your volumes - we will prepare a commercial proposal.