Welcome to CustodianAI
CustodianAI gives you two tools for building privacy-safe AI applications: a Python SDK for creating and deploying AI assistants, and Privacy Data Transformation for de-identifying sensitive text and documents before they reach your models or data stores.
What's in the docs
Python SDK
Build and deploy conversational AI assistants in a few lines of Python. The SDK handles assistant creation, knowledge file uploads, RAG-based retrieval, and multi-turn chat sessions — all through a single API key.
Privacy Data Transformation
Detect and remove PII and proprietary sensitive information from text and documents. Use it as a standalone API or enable it inside the SDK to automatically mask sensitive content before it reaches the LLM. Includes Guardian Layer — CustodianAI's proprietary detection mode for domain-specific sensitive content.
→ Get started with Privacy Data Transformation
New here?
Start with the Introduction for background on what PII de-identification is and when you need it, then follow the Quickstart to make your first API call.
Plans & credits
Every Privacy Data Transformation API call consumes character credits based on the size of the text processed. The Python SDK does not consume credits directly — credits are charged by the underlying API calls if PII masking is enabled.
| Plan | Character Credits | Best for |
|---|---|---|
| Free | 50,000 | Trying things out |
| Starter | 10,000 | Small projects |
| Professional | 100,000 | Production workloads |
| Beta | 500,000 | Early-access high volume |