Jak Voice AI przetwarza dane konta, produktu, glosu, czatu, rozliczen i bezpieczenstwa.
Operational legal template. Review and approve with qualified legal counsel before production use.
This template describes the intended operational privacy posture for the Voice AI application. It covers public website visitors, customers, tenant users, dashboard users, and end users who interact with customer-configured voice or chat assistants.
The customer remains responsible for telling their own end users how the customer's assistant is deployed and what customer-specific data is collected.
Account data includes name, email, tenant identifiers, authentication state, preferences, and role assignments.
Product data includes assistant configuration, widget settings, knowledge spaces, uploaded or crawled content, conversations, transcripts, tool calls, and usage metrics.
Operational data includes security logs, audit records, request metadata, billing records, consent choices, and support correspondence.
AI processing may conditionally use Google Gemini, OpenAI, and Anthropic according to platform configuration. Data categories can include prompts, retrieved knowledge context, transcripts, embeddings input, images, PDFs, screenshots, and diagnostic metadata when those features are enabled.
We process account and product data to provide the service, secure accounts, maintain tenant isolation, troubleshoot incidents, calculate usage, and support customer requests.
Optional analytics and marketing cookies are used only where the user has consented. Essential cookies and security logs are used to operate and protect the service.
External AI providers may process content for text generation, vision extraction, embeddings, safety enforcement, abuse monitoring, and service reliability. Provider use is subject to configured policy gates, tenant isolation, redaction, and retention settings.
Conversation retention defaults to 30 days and can be configured within the product where available. Security and API logs use operational retention windows unless a shorter customer setting or legal hold applies.
Account deletion first marks tenant records as pending deletion, then a cron process performs final deletion after the configured grace period using the account deletion finalizer.
Users may request access, export, correction, restriction, portability, objection, withdrawal of consent, and deletion through the product privacy tools or support channel.
Where Voice AI acts as processor for customer-controlled assistant data, requests may need to be handled by the customer as controller.
| Data class | Examples | Default retention | Deletion trigger |
|---|---|---|---|
| Account and tenant profile | Tenant record, users, roles, contact email, company metadata | Active account lifetime plus deletion grace period | Account deletion cron after scheduled deletion date |
| Voice and chat conversations | Messages, transcripts, session history, assistant responses | 30 days by default, configurable 7-90 days | Cleanup cron retention window or account deletion |
| Knowledge spaces and RAG indexes | Uploaded documents, crawled pages, embeddings, Qdrant collections | Active space lifetime | Space deletion or account deletion finalizer |
| Security and audit logs | Login events, access logs, admin actions, privacy changes | 90 days operational default unless legal hold applies | Cleanup cron, anonymization, or final account deletion |
| Billing and tax records | Subscriptions, invoices, payment events, refunds, tax evidence | Statutory accounting period where required | Anonymization on erasure request; legal records retained as required |
| Cookie and consent ledger | Consent categories, policy versions, timestamp, request metadata | 12 months from consent event | Retention cleanup or account deletion where linked to tenant |