Your data
Page updated November 18, 2026. Provider terms checked November 18, 2026.
In short
Your database and files are hosted in Canada (Supabase, ca-central-1 region), which is outside Quebec under Law 25.
AI processing is done in the United States by Anthropic (Claude API). Anthropic's commercial terms prohibit training models on this data. Data sent is kept for up to 30 days, subject to exceptions set by the provider.
Emails (Resend) and payments (Stripe) are processed in the United States. The site is served by Cloudflare (global network, metadata in the United States).
Provider register
| Provider | Service | Data | Storage and processing | Outside Quebec | Source (checked Nov 18 2026) |
|---|---|---|---|---|---|
| Supabase | Database, files, sign-in | File data and uploaded documents | Canada (ca-central-1); some processing and support may occur outside Canada under its data processing agreement | Yes | Regions, DPA |
| Anthropic | AI: extraction, drafting, checking | Technical excerpts and interview answers, names replaced by roles, without salaries, SIN, birth dates, addresses or contact details | United States | Yes | Commercial terms, Retention |
| Cloudflare | Site, application, DNS | Requests in transit, technical logs; no client files stored | Global network, metadata in the United States | Yes | Data localization |
| Resend | Service emails | Email address, subject and link of notices | United States | Yes | DPA |
| Stripe | Payments (not yet active; invoices until then) | Name, email, payment data entered at Stripe | United States | Yes | Policy, DPA |
What the AI reads, and what it never receives
The AI reads technical excerpts of your documents and your interview answers, with role labels (for example "Person 1"). It never receives social insurance numbers, birth dates, addresses, salaries or personal contact details: these are removed by code before anything is sent. Excerpts are sent inside the request; no client file is stored with the AI provider. Names are put back only in your package, by code, on the T661 lines that ask for them.
Training and retention
We train no model on your data. Under its commercial terms (section B), Anthropic "may not train models on Customer Content from Services". Anthropic deletes API inputs and outputs within 30 days, subject to exceptions it sets (for example content flagged for a policy violation, or a legal obligation). Northline intends to request zero data retention; this page will be updated with the answer.
Law 25
Person in charge: Yahia Ouardi, founder, hello@northlinerd.com. Privacy impact assessment: draft in progress. Written agreements: Northline relies on the data processing agreements published by each provider; the review of these agreements against sections 17 and 18.3 is in progress. Privacy incident register: in place. You can request access, correction or deletion of your data at any time.
Deletion
An account owner can delete their organization from the platform. Deletion erases the database rows and stored files.
Certifications
Northline holds no security certification (such as SOC 2 or ISO 27001). Our providers' certifications belong to them.