Privacy Policy

Last Updated: October 2026

1. Information We Collect

From Creators (Account Holders): We collect your email address, authentication data, IP addresses, and the robust biographical/psychological training data you actively submit to the Knowledge Base in order to shape your Neural Clone.

From Visitors: We collect persistent anonymous tracking identifiers (e.g., `visitorId`), connection metadata (IP address, Browser Type), and the contents of the conversations you maintain with a Creator's AI Clone. If you submit a Priority Request, we collect your provided Name and Contact Handle.

2. How We Process Conversation Data

The core feature of Avla involves semantic analysis and Generative AI. To provide this service:

  • Chat messages are processed in real-time by third-party Subprocessors (OpenAI APIs) to generate contextual responses.
  • We operate under strict Zero-Data-Retention agreements with our Subprocessors, meaning OpenAI does not use your private conversations to train their foundational models (e.g., GPT-4).
  • Raw chat logs are stored in our secure, encrypted database (Supabase/PostgreSQL) to power the Creator's Ego Analytics dashboard and Inbox features.

3. The "Decompiler" Feature & Raw Data Disclosure

As a Visitor, you acknowledge that Creators voluntarily upload their private truths to the database. By using the "Decompile Memory" premium feature, Visitors are simply unlocking access to data that the Creator explicitly authorized for revelation under the paywall parameters. Avla does not expose unauthorized, deleted, or unlinked private data.

4. Security and Data Retention

We implement enterprise-grade encryption (TLS/SSL in transit, AES-256 at rest) via our cloud providers. We retain chat logs and analytics data indefinitely to train the user's clone, unless the Creator explicitly requests account deletion (Right to be Forgotten under GDPR/CCPA). If a Creator deletes their account, all associated Knowledge Base records, reasoning answers, and Visitor conversation logs associated with that profile are fundamentally and irreversibly destroyed via cascading deletion arrays.