Privacy policy · Clovie

1. What we collect

Account info: Email, name, organization name, and role — provided during signup and team invitations.
Agent traffic metadata: Operational metadata for each gateway request — agent identity, model used, token counts, timing, cost, and policy decisions. Prompt and response content passes through the gateway in transit but is not stored in our database.
Trace data: If you configure trace storage, trace data (including prompt and response content) is sent to your trace project using your credentials. We do not store trace content in our infrastructure.
Governance data: Policy definitions, policy decisions, approval requests and decisions, budget configurations, anomaly detection records, and compliance export history.
Usage data: Page views, feature usage, performance telemetry, and session information to improve the product.
Audit & security logs: Sign-ins, administrative actions, access denials, kill switch activations, and policy changes.

2. How we use data

We use data solely to provide, secure, and improve the service:

Authenticating and authorizing AI agent requests via virtual keys
Evaluating policy decisions and enforcing budget controls on agent traffic
Generating cost attribution reports and chargeback data
Producing compliance evidence packs and audit trails
Detecting anomalous agent behavior and shadow (unregistered) agents
Smart model routing to optimize cost and quality

We do not sell your data. We do not use customer data to train AI models. We do not share your data with other customers.

3. AI processing

AI agent gateway

Your AI agents' requests pass through our OpenAI-compatible gateway to LLM providers (OpenAI, Anthropic, Google, and others). We record operational metadata (model, tokens, cost, timing, policy decisions) but do not store prompt or response content in our database. The gateway acts as a transparent proxy — your agents receive responses directly from the LLM provider via our infrastructure.

Smart model routing

When smart routing is enabled, Clovie analyzes request complexity to select the most cost-effective model. This analysis happens inline and is not stored beyond the routing decision metadata.

Your governance settings

Control which policies govern agent behavior, which models agents can access, and budget limits per agent and team.

4. Multi-tenant isolation

Customer data is strictly isolated in the database itself, so the isolation applies even if the application has a bug. This is supplemented by a second check in the application. Virtual keys, agent records, policy decisions, cost data, and all governance records are tenant-scoped.

Gateway operational data (runs, spans, costs, policy decisions) is always stored with the organization it belongs to and the same database-level isolation.

5. Data retention

We retain operational data for as long as your account is active. Policy decisions and approval records are stored in append-only tables and retained for compliance purposes.

On account deletion, customer data is soft-deleted immediately and hard-deleted within 30 days. Audit logs are retained for the period required by applicable regulations.

You may request data export at any time via the compliance export feature or by contacting us.

6. Your rights

You may request export or deletion of your personal data at any time via the in-product compliance export or by emailing hello@clovie.io. We comply with GDPR, CCPA, and similar data protection regulations applicable to your jurisdiction.

For AI agent data specifically, you control:

Which agents are registered and active
What budget limits and policies apply
Whether trace data is sent to your trace storage
When to export or delete governance records

7. Sub-processors

We use the following vendors to operate the service:

Vendor	Purpose	Data processed
Google Cloud Platform	Infrastructure, compute, database, secrets	All service data
OpenAI, Anthropic, Google	LLM providers (via gateway)	Agent prompts and responses (in transit)
Trace storage provider	Trace storage (customer-configured)	Agent trace data (if enabled)
Identity provider	Authentication and SSO	User identity, email, login events
Redis caching layer	Caching	Cached agent metadata, session data (ephemeral)

We do not share your data with any vendors beyond those listed above. An updated list is available on request.

8. Security

We follow industry best practices for security:

Encryption in transit (TLS 1.2+ on all endpoints) and at rest (AES-256)
An additional layer of encryption for sensitive credentials
Secrets kept in a managed secret store, never in code or environment files
Each organization's data isolated at the database layer
Role-based access control with seven role tiers
Append-only audit logging for all policy and approval decisions
Agent keys stored only as one-way hashes
Authenticated communication between our services

Report security issues to hello@clovie.io. We commit to acknowledging reports within 24 hours.

9. Children

The service is not directed to anyone under 16 and we do not knowingly collect personal data from children.

10. Changes

We may update this policy from time to time. Material changes will be communicated via email or in-product notification. Continued use of the service after changes constitutes acceptance.

11. Contact

Questions about this policy or your data? Email hello@clovie.io.