OPERATOR 91Security
Operator 91 holds the data that runs a transport company — drivers, rates, pay, and safety records. We designed the platform so that protecting that data is a property of how it's built, not a policy stapled on afterward. Here is what we actually do, in plain terms.
Tenant isolation
Every record belongs to exactly one customer, and access is scoped to that customer on every query. One customer cannot read or reach another's data.
Isolation isn't left to reviewer memory — automated invariant checks and an end-to-end test suite assert the tenant boundary on every change before it can ship.
New data models must be registered for tenant scoping or the build fails. The default is isolation; opting out takes a deliberate, reviewed exception.
Access & role gates
- Role-based access. Owners, managers, dispatchers, and drivers each see only what their role allows. Access to a person's records is separate from having a login — an admin grants it explicitly.
- Hard gates on the sensitive stuff. Money (pay, settlements, rate dollars, budgets), HR, and safety data sit behind role checks enforced on the server, never merely hidden in the interface. Widening someone's access means changing their role, not exposing data piecemeal.
- Two-factor authentication is required for admin and management accounts and available to everyone.
Sensitive data & money
Government identifiers are never stored in the clear. Where the platform needs an identifier like an SSN only to match a person, it stores a one-way keyed hash plus the last four digits for display. The raw number is never persisted, logged, or returned by the platform.
We never move money. Financial, factoring, fuel-card, and settlement data is a read-only mirror. The platform prepares files a human takes to their own accounting system; it initiates no payments, transfers, or payroll. There is no code path that moves funds.
Advisory AI, on a leash
- Propose, never act. Operators (our AI) read a customer's data and queue suggested actions for approval. A person confirms every consequential change; the AI cannot execute on its own.
- Validated twice. Every proposed action is schema-validated when it's proposed and again when it's executed, and the executed record names the human who approved it.
- Scoped and private. AI runs against a customer's own data using the provider key that customer configures. We do not use customer operational data to train general-purpose models.
Application & infrastructure
- Encryption in transit. All traffic is served over HTTPS with HSTS; this site is delivered from a hardened static host.
- Security headers & CSP. This website enforces a strict Content-Security-Policy (no third-party scripts), plus X-Content-Type-Options, X-Frame-Options: DENY, a restrictive Referrer-Policy, and a locked-down Permissions-Policy.
- Encrypted secrets. Credentials for connected tools and AI keys are stored encrypted; the platform surfaces only whether a key is set, never the key itself.
- Guarded public links. No-login surfaces (a customer's tracking window, an applicant's link) use short, purpose-scoped tokens, run behind rate limits, set tenant context explicitly, and expose only outsider-safe fields — with a version the company can bump to revoke a link instantly.
- Managed, backed-up hosting. The application runs on managed cloud infrastructure with automated database backups; deploys are gated on the full test suite passing.
Data ownership & lifecycle
Your data is yours. You can export your core records as CSV at any time, and request deletion of your account data (subject to records we must retain by law). What handling of personal information looks like in detail is in our Privacy Policy.
What we don't claim (yet). We believe in being straight about our stage. Operator 91 does not currently hold a SOC 2 or ISO 27001 certification — we're an early-stage platform, and we'll pursue formal attestation as we grow and as customers need it. What we can show you today is the substance above: isolation and gates that are tested on every change, not just described in a slide.
If you're evaluating us and need to go deeper — a security questionnaire, our subprocessor list, or a data-processing addendum — just ask.
Reporting a vulnerability
Found something? We want to hear from you. Email [email protected] with the details and steps to reproduce. Please give us a reasonable chance to fix an issue before disclosing it publicly; we won't pursue good-faith researchers who follow that path.