Security
Loreflow stores the most sensitive thing a company owns — its reasoning. This page is maintained by the Loreflow team to explain, in plain terms, the security model that data is handled under. Loreflow is being built to this model; for the current status of any individual control or certification, ask us.
Data is encrypted in transit and at rest, with keys rotated automatically and isolated per workspace.
Loreflow is designed to mirror the access rules of every source it reads. If someone cannot open a document in the source tool, it should not exist for them in Loreflow.
Original content stays in the source tool. Loreflow keeps references, relationships and summaries, and fetches the original on demand with the viewer's own permissions.
Access to production is limited, logged and reviewed. Every administrative action inside a workspace is recorded.
Loreflow runs a SOC 2 Type II program and supports GDPR obligations including DPAs and subprocessor disclosure. Certification status, current subprocessor list and penetration test summaries are shared under NDA on request — ask us rather than assuming a status from this page.
Send findings to security@loreflow.example. We acknowledge reports within one business day and will keep you updated through remediation.
Posture at a glance
Sample figures illustrating the targets the production platform is held to. Ask us for the current measured position, the report, the policy or the log.
0.00%
Uptime, trailing 90 days
API and sync workers measured together.
0 hrs
Critical patch window
Median time from disclosure to production deploy.
0
Customer data incidents
Since the company was founded. Incident history is public.
0%
Encrypted at rest and in transit
AES-256 at rest, TLS 1.3 in transit, no exceptions.
Loreflow is built to read your sources with the same access model they already enforce. If a document is invisible to someone in Notion, it stays invisible to them here — including inside search results, answers and the graph.
Independent annual penetration testing is part of the security program; report available under NDA.
Access decisions per query
Every answer is filtered per person before it renders.
−97% from previous step
−30% from previous step
−100% from previous step
How data flows
Read from source
Read-only OAuth scopes, scoped to what you tick. We never request write access to a source unless a feature you enabled requires it.
Process and link
Content is encrypted in transit and at rest, processed in an isolated tenant, and never used to train a shared model.
Serve with filters
Answers are assembled per user, per query, with the source permission checked at the moment of the request.
Plan differences
Security fundamentals are included on every plan. Governance and compliance tooling scales with the tier.
| Capability | Pro Team | Startup | Enterprise |
|---|---|---|---|
| Encryption at rest and in transit | ✓ | ✓ | ✓ |
| Permission mirroring | ✓ | ✓ | ✓ |
| Audit log | 1 year | Unlimited | Unlimited + legal hold |
| Role & scope permissions | Standard | Granular | Granular |
| Scoped investor access levels | 1 seat | Unlimited | Unlimited |
| SSO / SAML | — | — | ✓ |
| SCIM provisioning | — | — | ✓ |
| Data residency choice | — | — | ✓ |
| Private AI models / BYO key | — | — | ✓ |
| Signed DPA and sub-processor alerts | ✓ | ✓ | ✓ |
We answer reviews, DPAs and procurement forms within one business day, and we will happily join your vendor call.
In an isolated per-workspace store, encrypted at rest and in transit, with regional hosting options for EU-only requirements.
Read-only scopes per connector, narrowed to the resources you select. There is no write access to any source system.
No. Workspace content is never used to train models, and AI recall runs against your workspace only.
Yes. Questionnaires, DPAs and subprocessor lists are answered within one business day — see the DPA and subprocessors pages.
Role-based access with audit logging on every read. Support access to a workspace is time-boxed and requires explicit approval.