Data handling and subprocessors
What happens to a document after you upload it, who processes it and where. Last reviewed October 2026.
How uploads are processed
Parsing and structuring
Word, Excel and PowerPoint files are parsed on request to extract text. The parsing endpoint returns the text to your session and does not write files to a document store.
Executive assessments
Assessments are scored in the application from your structured evidence and cite the source section for each finding. They do not call a third-party AI model.
AI-assisted summary and chat
These features send text to a model provider through Vercel AI Gateway. API terms for that provider exclude customer data from model training by default. Do not upload content you are not permitted to process with a subprocessor.
Subprocessors
| Subprocessor | Purpose | Data involved | Hosting region |
|---|---|---|---|
| Vercel | Application hosting and AI Gateway routing | Requests, uploaded file contents in transit | Region set per Vercel project; confirmed in the DPA on request |
| OpenAI (via Vercel AI Gateway) | AI-assisted document summary and the Assist chat | Up to 15,000 characters of extracted document text per analysis; chat messages | Provider-managed; confirmed in the DPA on request |
| Neon | Managed Postgres database | Account and application records | EU (Frankfurt, eu-central-1) |
| Supabase | Authentication | Email address and account identifiers | Confirmed in the DPA on request |
| Stripe | Payments and subscriptions | Billing details handled by Stripe | Stripe-managed |
| Resend / Hostinger email | Transactional email | Name and email address | Provider-managed |
Procurement or legal teams can request a data processing agreement and the current region list at cosimo@aliverogs.io. See also our privacy policy.