Privacy, in plain language.
This document specifies the temporary in-memory processing, strict tenant isolation boundaries, and automated cryptographic deletion protocols governing Clarivect.
Automatic scheduled cleanup removes source files and scan records after 24 hours.
Execute “Delete Now” at any stage to immediately purge all raw payloads and derived findings.
Non-public memory processing with TLS 1.3 transport encryption and AES-256 token storage.
Uploaded corporate data is never ingested into LLMs or used to train external foundation models.
What the scanner processes
The scanner processes only the CSV files you select, mapped invoice and counterparty fields, validation errors, deterministic findings, feedback you optionally provide, and the generated Receivables Assurance Brief.
- Invoice metadata (identifier, issue date, due date, amount, currency)
- Debtor identifier or counterparty naming for deduplication and grouping
- Payment terms and payment timestamp records where present
- User-confirmed column mapping specifications
Scan expiry and retention
Session access expires automatically after 24 hours. Scheduled cleanup routines remove expired source files and temporary scan records. Access expiry and physical cleanup are separate deterministic steps.
Immediate deletion protocol
Clicking “Delete Now” immediately executes an unrecoverable hard-deletion job:
- Destroys the uploaded raw CSV and column mapping cache
- Purges temporary invoice and customer entity snapshots
- Erases row-repair diffs, findings, rule evaluations, and generated PDF briefs
- Removes temporary feedback and profiling responses
Zero AI model training
Uploaded CSV data is never used to train public or proprietary foundation models. The scanner's financial calculations, aging calculations, duplicate detections, and priority scores are executed using deterministic mathematical rules—not probabilistic AI models.
Where commercial AI extraction models are used for complex document parsing (in enterprise services), payloads operate exclusively under zero-data-retention (ZDR) commercial agreements with immediate memory discard.
Optional feedback and profiling
Feedback and business-profile questions appear only after results are computed and are entirely optional. They never block report download or deletion. Providing product feedback, granting permission to contact you, and granting permission to publish approved testimonials are handled as distinct, unbundled choices.
Product interest and site analytics
You may separately provide an email address and consent to receive updates regarding Clarivect software. That request does not retain your uploaded financial data. Privacy-preserving site analytics record limited funnel events (such as page views, scan initiation, and report download counts); they do not capture invoice values, customer names, uploaded files, or scan findings.
Workspace continuation & scan lineage preservation
An anonymous scan does not create an ongoing receivables workspace. If you sign in and explicitly choose to link a completed scan, data is segregated according to this strict architectural boundary:
- Scan reference token & conversion timestamp
- Account and workspace foreign key linkage
- Ruleset version and financial calculation metadata
- Explicit operator consent verification audit log
- Raw CSV file payloads and column mappings
- Individual invoice numbers, line-items, and debtor names
- Row-repair data and derived PDF brief binaries
- Customer and debtor specific aging breakdowns
Security, governance and contact
Scanner data is isolated in non-public storage and accessed through an expiring private session. Current hosting, subprocessor registries, and vulnerability disclosure policies are published in our Trust & Security center.