Trust Center
What we do with your data
Everything on this page is available to every plan, without a sales call. Where something is still in progress we say so — a compliance page that overstates itself fails the exact review it exists to pass.
Security
- Encryption
- TLS 1.2+ in transit; AES-256 at rest for the application database, object storage and backups.
- Access control
- Org-scoped roles (owner, admin, member, viewer). API keys are scoped per resource and can carry their own run and spend budgets, so an integration can never spend more than you minted it for.
- Tenant isolation
- Every workspace is its own Postgres schema, and the search path is set per transaction — isolation is enforced by the database, not by a WHERE clause someone has to remember.
- SOC 2 Type IIIn progress
- Audit in progress. We will publish the report here and share it under NDA on request — we would rather tell you the date than imply a certificate we do not yet hold.
- Penetration testingIn progress
- Annual third-party test; summary letter available on request.
Your data
- We do not train on your data
- Your Blueprints, research and exports are never used to train models — ours or a provider's. DeepSieve sends prompts to model providers under zero-retention terms where the provider offers them.
- You own it, and you can leave with it
- Bulk export in NDJSON, JSON or CSV at any time, with citations intact. If you cancel, your data stays readable and exportable through the retention window — cancelling is not deleting.
- Deletion
- Delete a Blueprint and its schema goes with it. Account deletion purges tenant data after the retention window; backups age out on their own schedule.
- Provenance
- Every cell carries a verdict, including “not checked”. The ones we could source carry their source URLs, and the ones we scored carry a confidence number too. Where a value could not be verified it says so rather than being passed off as sourced — so what you publish downstream can be traced back, and what can't be is visible before you publish it.
Infrastructure & residency
- Hosting
- AWS, in private subnets behind a WAF, with managed Postgres and automated backups. No customer data on developer machines.
- Data residency
- US by default. Regional hosting (EU) is available on Enterprise — ask before you sign, not after.
- Availability
- Enterprise plans carry a written SLA. Below that we publish incidents rather than quoting a number we have not committed to contractually.
Legal & compliance
- DPA
- A click-through Data Processing Agreement is included on Pro and Enterprise (negotiated terms on Enterprise). On Starter, ask and we will send one — you do not need a sales call either way.
- Subprocessors
- We publish the list below and notify customers before adding one.
- GDPR / CCPA
- Data-subject requests are handled through your account owner; contact us and we will action them within the statutory window.
- Security review
- Enterprise includes a full security review — questionnaires, architecture walkthrough, and the pen-test summary.
Subprocessors
Third parties that may process customer data. We notify customers before adding one.
| Provider | Purpose | Region |
|---|---|---|
| Amazon Web Services | Hosting, database, object storage | US |
| Google (Gemini) | Research + extraction models | US |
| Brave Search | Public-web search — receives your research queries | US |
| Anthropic / AWS Bedrock | Model inference (deployment-dependent) | US |
| WorkOS | Authentication, SSO, directory sync | US |
| Stripe | Payment processing | US |
| Metronome | Usage rating and invoicing | US |
| Resend | Transactional email | US |
| Langfuse | LLM observability | US / EU |
Security questionnaire, DPA or pen-test summary? support+deepsieve@honeynudger.ai