Where your data is processed
Last updated: July 26, 2026
We believe in being clear about where your data lives and where it is processed. Here is the distinction, with no ambiguity.
Your data at rest stays in Canada
- Documents (receipts, invoices, statements): stored on Google Cloud Storage, Montréal region (`northamerica-northeast1`).
- Encrypted backups: a dedicated Canadian bucket (`dossiera-backups-ca`).
- Database: hosted in Canada.
- Inbound-email reception buffer: AWS Canadian region (`ca-central-1`).
This does not change. It is a guarantee we continue to make.
AI processing may be routed outside Canada
AI extraction (reading a document to pull out vendor, amounts and taxes) runs on Claude models. The request is issued from Canada (`ca-central-1`), but the compute may run outside Canada (United States) by our declared providers:
- Amazon Bedrock — cross-region inference profile (United States / global).
- Anthropic API — United States (used as a fallback when Bedrock capacity is saturated).
The AI providers do not retain your documents beyond processing the request, and this data is not used to train their models.
Why we made this choice
A "processing in Canada only" constraint locked the product onto an end-of-life generation of AI models. Relaxing it gives us access to the latest models — more accurate extraction — while keeping your documents at rest in Canada. The extraction result remains suggestive: it is reviewed and validated by a human (the firm) before any accounting processing.
Your recourse
For any question about how your information is processed, or to exercise your rights (Law 25 / PIPEDA), write to privacy@dossiera.ca. See also our privacy policy and the sub-processor register it contains.