Skip to content

Security, privacy & compliance

Use this section to evaluate Venturi’s trust boundary, access model, data handling, availability controls, and supporting evidence before deployment.

This section is written for the people who review Venturi before it is deployed: your security team, your privacy and data-protection office, your procurement reviewers, and (where it applies) your EU works council. Every claim here is grounded in the platform’s binding architecture and its published controls.

The security posture in one page

  1. Venturi runs inside a dedicated tenant boundary. Self-hosted mode deploys in your cloud account; SaaS mode uses a Venturi-operated single-tenant environment in your contracted residency lane. Operational data stays inside that boundary. The separate control plane receives no raw tenant data.
  2. Every integration is read-only. Connectors request read-only scopes wherever the provider offers them, the IAM policies attached to Venturi’s workloads carry an explicit Deny on write actions against your source systems and cloud resources, and the customer cross-account role grants no write permission. Provider-account connectors use dedicated credentials with the narrowest available provider permissions and send only GET requests to allowlisted read endpoints. OpenAI organization Admin keys support permission selection; the provider matrix separates them from Anthropic reporting credentials. Venturi never writes to a system it reads from.
  3. A failing Venturi dependency never blocks your AI traffic. The decision-time interceptor is fail-open within a 50 ms P99 end-to-end budget, bounded by a cooperative timeout against a cumulative deadline: if any Venturi dependency behind it is slow, degraded, or unavailable, your request is forwarded unmodified. Fail-open is not configurable. With the drop-in proxy, the proxy endpoint’s own reachability stays on your request path; see Proxy endpoint reachability.
  4. Outputs are confidence-bounded so financial decisions never rest on false precision. Operational confidence is capped at 0.95; the Chargeback-ready band (coper >= 0.80) is a necessary confidence floor, and export_eligibility, dispute, reconciliation, and finance-acceptance gates also apply.
  5. Adoption intelligence is cohort-only by construction. Every cohort read path enforces a server-side minimum cohort of 5. Venturi performs no individual-worker profiling and no emotion or behavioral-state inference.
  6. No prompt or completion content is ever stored. Venturi attributes consumption from metadata, content-bearing provider log records are filtered in memory before anything is materialized, and content on a call routed through the drop-in proxy exists only in memory while it is forwarded.

How to read this section

The pages below describe Venturi’s security controls. Your implementation checklist records tenant-specific configuration and verification evidence. Attestation status is stated factually: SOC 2 Type I readiness/report target on our security roadmap with Type II targeted within 12 months; it is not yet held. Venturi does not hold a SOC 2 report, ISO/IEC 27001 or ISO/IEC 42001 certification, or a FedRAMP authorization. The GDPR processor artifacts are included in the onboarding diligence pack. Deployment-specific EU AI Act assessment and applicable market-entry obligations are completed before EU market placement or putting into service. See Compliance for each item.

Page What it answers
Security architecture The security design: trust boundary, read-only integrations, encryption, secrets, fail-open behavior, and threat model.
Tenant isolation How your data is isolated from every other tenant: per-tenant stores and keys, cross-tenant rejection, and break-glass support access.
Data privacy & retention What Venturi does and does not collect, the 13-month retention default, and crypto-shred erasure.
Data-subject rights How access, erasure, and portability requests are handled, and Venturi’s processor role.
Residency & subprocessors Where data lives, region pinning, and where to find the versioned subprocessor list.
Compliance Covers SOC 2 readiness, penetration testing, GDPR posture, and CCPA controls. Documents the deployment-specific EU AI Act assessment and market-entry obligations.
Trust center The consolidated trust summary, the control framework crosswalk, the onboarding diligence pack, and the vulnerability disclosure program.
Technical diligence A focused review of prompt and completion non-retention, read-only connector controls, tenant isolation, and SOC 2 status.
Complete the trust-center review The in-product steps a security reviewer follows to review and acknowledge the trust checklist.

The invariants behind everything

A small set of frozen invariants shapes every security control in this section. They hold on every deployment and cannot be configured away.

Fail-open protects requests that reach the interceptor

Failures or timeouts in Venturi dependencies behind the decision-time interceptor forward admitted requests unmodified. The drop-in proxy endpoint’s own reachability remains on your request path; see Proxy endpoint reachability. The synchronous decision-time interceptor works to a 50 ms P99 end-to-end budget, bounded by a cooperative timeout against a cumulative deadline, and fails open on breach: your traffic is forwarded unmodified and the event is logged with null attribution. Authentication, authorization, tenant isolation, export, and billing fail closed; admitted customer traffic never does. The one deliberate rejection is the drop-in proxy’s tenant admission: a call that carries no provider key, or a key that is not registered to your tenant, is rejected before anything is forwarded (see Tenant admission). See Security architecture.

Read-only integrations, enforced in layers

Every connector is granted read-only vendor permissions (scopes, roles, or grants) wherever the vendor offers them. Where a vendor scope is broader than read-only, such as Salesforce api, the integration user’s permissions enforce the read-only boundary. Where the credential permits more than the required reads, the connector guide names that capability and the connector’s independent endpoint allowlist. The provider matrix distinguishes Anthropic reporting credentials from OpenAI’s selectable read permissions. The IAM policies attached to Venturi’s workloads contain an explicit deny on write actions against your source systems and cloud resources, and the customer cross-account role grants no write permission. This is a frozen invariant: no flag or connector setting enables a write.

No content capture

Content capture is structural, not configurable: no setting turns it on. Venturi processes invocation metadata (identity, service, project, cost, tokens, timing), never prompt or completion bodies. Where a provider log source can contain content, the collector discards content fields in memory and never persists them; see Content-bearing logs. Provider API keys on proxied calls are never stored: Venturi keeps only a one-way keyed fingerprint and a truncated display prefix (see Tenant admission).

Bounded confidence, never false certainty

Operational confidence (coper) is policy-capped at 0.95 and cannot be raised. The Chargeback-ready band (coper >= 0.80) is a necessary confidence floor. Export eligibility (export_eligibility=eligible) also requires eligible output state, current evidence, and no open dispute or degraded result; reconciliation and finance acceptance govern publication. A score does not establish financial accuracy. Unknown cost is reported as unknown, never as zero. See the confidence model.

Where to start

Two other pages explain the architecture outside this section, at different depths:

  • How Venturi works is the short conceptual overview of what Venturi observes, attributes, and reports.
  • System architecture is the engineering reference for plane topology, the synchronous and asynchronous paths, and failure boundaries.

Security architecture, in this section, covers the security controls that sit on top of that system design.

Give feedback on this page

Sign in to send feedback. Opens in a new tab.

Copy page details 
Docs feedback: Security, privacy & compliance
Page: https://docs.venturi.systems/security/