Jatter is built on a simple principle: your data belongs to you. This isn’t just a policy—it’s enforced through architecture. This page explains, in detail, how your data is handled, secured, and protected at every step.
Core Privacy Guarantees
1. Client-Side Encryption by Default
All sensitive user data is encrypted before it leaves your device.
  • Learned web page data is client-side encrypted
  • Chat history is client-side encrypted
This means Jatter’s servers store only ciphertext, not readable data.
2. Zero Data Retention for Personal Reasoning
When Jatter generates answers using your personal data:
  • Data is decrypted only ephemerally
  • It is processed in-memory for retrieval and reasoning
  • It is never stored, logged, or retained
Once the response is generated, your data is gone from the server.
3. Zero Data Retention for Grounded Open Web Reasoning
Jatter can generate answers using information from the open web through a trusted third-party reasoning service. This capability is designed to remain fully ephemeral and privacy-preserving.
When grounded reasoning is used:
  1. Your query is sent to a third-party service configured under an enterprise zero data retention agreement
  2. The service retrieves and processes relevant public web content
  3. The response is returned to Jatter without any data being stored or retained
No web content, queries, or intermediate data are persisted by Jatter or the third-party provider.
4. Your Data Is Never Used to Train Models
Jatter enforces strict data isolation:
  • Your data is never used to train AI models
  • Your data is never sold or shared
  • Your data is used only to serve your requests
5. Secure Infrastructure by Design
  • All data is encrypted at rest
  • All data is transmitted over HTTPS (TLS)
  • Systems are designed to minimize data exposure at every layer
6. Controlled Logging for Abuse Prevention
To maintain system integrity:
  • Chats sent to our LLM service are anonymized
  • These logs are retained for up to 55 days
  • Logs are used only for abuse detection and system safety
No personal identity or encrypted user data is exposed in this process.
How Jatter Handles Your Data
Learning Web Pages
When you choose to let Jatter learn from a webpage:
  1. The page text is extracted locally
  2. The text is sent securely over HTTPS to Jatter’s servers
  3. The server:
    • Processes the text
    • Splits it into chunks
    • Generates embeddings
  4. The original text is not retained
  5. The chunked data is:
    • Encrypted on your device
    • Sent back to Jatter’s servers as ciphertext for storage
Result:
Jatter can retrieve relevant knowledge later—without ever storing readable versions of your data.
Chat Processing
When you interact with Jatter:
  1. Your chat is stored encrypted on your device and in the cloud
  2. When you send a message:
    • Relevant chat context is decrypted temporarily
    • Sent to the LLM for processing
  3. After the response:
    • The data is immediately re-encrypted
Accessing Learned Data (Personal Answers)
When Jatter answers questions about your personal data:
  1. Relevant encrypted data is retrieved
  2. It is decrypted only for the duration of the request
  3. The system performs retrieval-augmented generation (RAG)
  4. Once the response is generated:
    • All decrypted data is discarded
    • Nothing is stored or logged
This is what enables powerful personal answers—without persistent exposure.
What This Means in Practice
  • Even if someone accessed Jatter’s storage systems, they would see only encrypted data
  • Your browsing-derived knowledge is not readable by Jatter
  • Your conversations remain private and inaccessible
  • AI responses are generated without creating lasting copies of your data
Why This Matters
Most AI systems require you to trust that your data won’t be misused.
Jatter removes that requirement.
Through client-side encryption, zero-retention processing, and strict data boundaries, Jatter ensures:

Your data is not just protected by policy—it is protected by design.