Textql Trust Center

generated: '2026-08-30'
method: searched
source: https://textql.com/security
url: https://trust.textql.com/
trust_center:
  url: https://trust.textql.com/
  platform: Drata
  redirects_to: https://app.drata.com/trust/36f9232d-d1fa-4cc8-ac65-746a94a71cc3
  probed: '2026-08-30'
  http_status: 403
  readable: false
  note: >-
    The trust center is linked from textql.com/security as "Visit Trust Center" and resolves to a
    Drata-hosted portal, but returned 403 to both a browser-User-Agent curl and a WebFetch. Its
    contents could not be read this run. The certifications below are therefore taken from TextQL's
    OWN security page, which was fetched successfully, rather than from the portal.
certifications:
- name: SOC 2 Type II
  source: https://textql.com/security
  detail: Audited annually, report on request.
- name: HIPAA
  source: https://textql.com/security
  detail: BAAs signed; PHI never leaves the customer's cloud. Enterprise tier.
- name: GDPR
  source: https://textql.com/security
  detail: DPA available, EU data residency.
- name: SOX
  source: https://textql.com/pricing
  detail: >-
    Listed in the Enterprise tier's compliance column as "SOX, audit & policy support" — a support
    posture for the customer's own SOX programme, not a certification TextQL holds.
  qualifier: support-posture-not-certification
corrections:
  - >-
    An earlier automated pass recorded ISO 27001 from a keyword match on the Drata portal URL. That
    claim is REMOVED — the portal returned 403 and was never read, and ISO 27001 appears nowhere on
    TextQL's own security or pricing pages. Only SOC 2 Type II, HIPAA and GDPR are provider-stated.
security_posture:
  source: https://textql.com/security
  probed: '2026-08-30'
  http_status: 200
  model: >-
    "TextQL treats its own agent as an untrusted principal. Every query clears four gates — identity,
    entitlement, execution, audit — before a byte of your data moves."
  isolation: gVisor sandbox per session, destroyed when the session ends; no sharing between customers or between two users of the same customer
  encryption: TLS 1.2+ in transit, AES-256 at rest, customer-managed keys via customer KMS on VPC and on-prem installs
  deployment_options: [multi-tenant SaaS, single-tenant, customer VPC, BYOC, on-premises, air-gapped, private link]
  model_training: >-
    "Never trained on your data — not our models, not a provider's, with no contractual carve-out."
    Customers may bring their own Bedrock, Vertex or Azure OpenAI deployment so inference never leaves
    their account.
  identity: SAML 2.0 and OIDC (IdP- and SP-initiated), JIT provisioning, SCIM 2.0 lifecycle, domain claim enforcement, MFA enforced at the customer IdP
  authorization: >-
    Roles compose from named grants and resolve per request rather than at login. Row filters and
    column masks are evaluated by the customer's warehouse — Snowflake row access policies, Unity
    Catalog column masks, BigQuery policy tags — so revoking a grant upstream takes effect on the next
    query.
  network: gVisor-isolated sandbox per run, outbound domain allowlist, credential-injecting egress proxy, PrivateLink/VPC peering/static egress IPs on enterprise
  secrets: Credentials live in a vault the agent process cannot read.
  audit: Audit Log (AuditLogService in the Public RPC API) covers security and administrative activity