Third-Party & LLM Data Practices
Status
Owner: Tech lead · Status: draft · Canonical source: flezi.app.ai.agent/app/llm/
Provides full transparency regarding external services receiving customer prompts, data residency regions, and model training policies.
Sub-processor Details
| Service | Provider | Region | Data Processed | Model Training Policy |
|---|---|---|---|---|
| Azure OpenAI Service | Microsoft Azure | Southeast Asia / East US | Prompts, completion context | Microsoft guarantees customer data is NOT used to train foundation models. |
| Anthropic API | Anthropic | US Central / Global | Prompts, queries | Commercial API terms strictly prohibit model training on API submissions. |
| Azure AI Search | Microsoft Azure | Southeast Asia | Embeddings, chunks | Isolated within customer enterprise tenant. |
Existing Controls
- All external API calls encrypted with TLS 1.3.
- User identity metadata (passwords, emails) is never passed into model prompt payloads.
Evidence (Files & Configuration)
flezi.app.ai.agent/app/llm/- Microsoft Enterprise Agreement & Azure DPA
Limitations & Roadmap
- Periodic review and re-certification of vendor DPA agreements.