TRUST

Single-tenant by default.
Yours, provably.

Every workspace is its own tenant — its own database, domain, and identity. QaiS reads only your data, and every action it takes is logged, attributable, and reversible.

All systems operational · in production since 2024
TRUST PILLARS

The guarantees under every workspace.

SOC 2 Type II

Independent attestation covering security, availability, and confidentiality — audited annually. Reports available under NDA.

Audited since 2024

Single-tenant data

Every workspace gets its own database, blob namespace, and identity scope. Cross-tenant queries are blocked at the database, not just the app.

Own DB · domain · keys

No cross-tenant training

QaiS never trains on your tenant data. Prompts and context stay scoped to the workspace and are not retained by the model provider beyond the request.

Your work teaches no one else

Every action audited

Every QaiS tool call and every user action is logged with the actor, a before/after diff, and one-click undo. Retention is plan-tiered.

Logged · attributable · reversible

Encrypted end to end

TLS 1.3 in transit and encrypted at rest. Prompts sent to LLM providers travel over TLS under data-processing agreements that forbid training.

In transit · at rest

Access control & SSO

Every request is authenticated and tenant-scoped. Role-based access across the workspace, with single sign-on on Business and Enterprise plans.

SSO on Business & Enterprise

Data residency & ownership

Run on the region you choose — US-East, US-West, EU-West (Frankfurt), UK (London), or APAC (Sydney) on Enterprise. The data is yours to export anytime.

Region of your choice

Reversible by design

Every action carries a before/after diff and a one-click revert. Full export in JSON and raw blobs on exit, then a verified hard-delete with a signed certificate.

One-click undo · clean exit
HOW THE AGENT IS GOVERNED

QaiS acts inside your tenant — with your rules.

QaiS runs inside your tenant, and only your tenant. It works under your team's rules — who can call which tool, and on whose behalf — and it can't reach another workspace's data.

Every call it makes is attributed to a person and a moment, carries a before/after diff, and can be reversed with one click. Nothing the agent does is off the record.

The log below is the same audit trail your admins see — every tool-call, timestamped, attributable, and reversible.

TENANT: ACME-FAB AUDIT TRAIL
09:14:06 QaiS projects.shift_milestone Frame · Cap 4 · +2d Reversible
09:14:07 QaiS tickets.create INSP-2241 · QA lead Reversible
09:14:09 QaiS messages.broadcast #ops-floor Reversible
09:15:32 a.martin finance.export Q2 forecast · PDF Reversible
COMPLIANCE REGISTER

On the record.

SOC 2 Type II Audited annually · since 2024
Single-tenant data Row-level isolation at the database
Every action audited Actor · diff · one-click undo
No cross-tenant training Prompts never retained for training
GDPR & CCPA — DPA available · EU SCCs · residency in Frankfurt
ISO 27001 — in progress · certification expected Q1 2027
Sub-processors — AWS · Anthropic · OpenAI · Stripe · Postmark · Cloudflare · Datadog

Run on a system you own.