Security
How Vedential isolates, protects and retains your data — tenancy, authentication, encryption, audit, backups, residency and AI data handling. For verification, due-diligence questionnaires or anything not covered here: security@vedential.com.
Data isolation
One isolated workspace per organisation. Every record carries the organisation's identity and the database enforces row-level security on every read — a query from your workspace physically cannot return another organisation's rows. This is enforced by the database itself, not by an application-layer filter.
Write enforcement
Browsers get read-only database access. Every create, update and delete goes through our server, which independently re-verifies identity, organisation and role authority for that specific action on that specific project before anything is written. Client-side checks are usability mirrors; the server is the authority.
Authentication
Email/password and Google sign-in. Passwords require a minimum of 10 characters including upper and lower case letters, numbers and symbols, and are checked against known-breached password lists at signup and reset. Password resets use single-use, any-device tokens that are only consumed on an explicit click (link scanners cannot burn them). Authentication endpoints are rate-limited. SAML single sign-on is an Enterprise capability provisioned per customer during onboarding (see Enterprise capabilities below); multi-factor authentication is on the roadmap.
Encryption
All traffic is encrypted in transit (TLS). Data is encrypted at rest on our database and file-storage infrastructure.
Permissions & roles
Role-based access with org-level roles (Admin, Portfolio Manager, Contributor, Viewer), a program-level Program Manager grant (full authority inside one program, nothing outside it) and per-project membership. Capability flags gate sensitive areas (budgets, risks, resources). Viewers are read-only by construction — write controls do not render for them, and the server refuses writes regardless. Administrators choose a seat policy (open, capped with one-click approval of seat requests, or IT-managed) and can export an access review — every member with role, grants, seat status and last sign-in — at any time.
Audit trail
Every mutation writes an audit entry — who, what, when. Audit attribution survives account deletion, and approval events on change requests are stamped server-side and cannot be supplied by a client.
Backups & recovery
Daily automated backups with a rolling 7-day window, plus manual snapshots before risky operations. Every migration ships with a documented rollback.
Data retention & deletion
Your data is yours: full workbook export is available at all times, on every plan, even after a subscription lapses. When you ask us to delete your organisation, we remove its records, files and user accounts from the live system; residual copies age out of the backup window within days. Deletion requests are honoured within 30 days, as stated in our Privacy Policy.
Data residency
Customer data is stored in Sydney, Australia. Optional AI processing (below) occurs outside Australia at the moment of each request; the AI provider deletes inputs and outputs within 30 days and is contractually barred from training on them.
AI data handling (optional features, preview)
AI features run only when a user explicitly clicks them — never in the background. Summary drafting sends only records the requesting user already has permission to read, assembled under the same row-level security as every other read. The plan builder sends only the description the user types. Launch-readiness reviews send only the documents the user ticks on that specific run; unticked files are never read, and document content is never stored or logged. Every AI output lands in a review step — nothing is written to your data until a human approves it, and approved output passes the same server-side validation as hand-entered data. The AI cannot invent monetary amounts, assign people, or write to the database; its output is checked by deterministic code, and we test this with adversarial (prompt-injection) cases before each AI release. Our AI provider does not train on customer data. Requests are audit-logged with usage counts only.
Subprocessors
We use a small set of infrastructure providers (hosting, database, email delivery, payments, AI processing). We disclose categories publicly and name specific vendors privately during customer security reviews.
Incident response
We maintain a documented breach-response plan and carry cyber liability insurance. Affected customers are notified without undue delay, consistent with our Privacy Policy and the Australian Notifiable Data Breaches scheme.
Reporting a vulnerability
Found something? Email security@vedential.com. We read every report and respond to genuine findings.
Continuity
Vedential is a small company, and we answer the continuity question directly. We maintain a written information-security management system aligned to ISO/IEC 27001 (scope, risk register, statement of applicability, policies and an evidence calendar; a management system, not a certification) and a business-continuity plan with scenario runbooks that we rehearse, including a full restore of the service from backup into a new environment. Your data is exportable in full at any time on every plan, so your portfolio is never held by the service. Both documents are available under NDA during a security review.
Enterprise capabilities
For organisations with advanced requirements, the Enterprise plan adds SAML single sign-on with identity-provider groups mapped to roles and programs, audit-log extracts on request, and — as separately priced add-ons — dedicated single-tenant instances for teams that require hard infrastructure isolation, and the option to run Vedential's AI features against your organisation's own approved AI model endpoint (so AI traffic stays under your existing AI agreement and data boundary). These are provisioned per customer during onboarding — scoped to your identity provider and evaluation checklist — rather than switched on self-serve. Talk to us about your requirements: sales@vedential.com.
Security questions, disclosure reports or questionnaire requests: sales@vedential.com.