Quickstart
Pick the path that fits what you're building.
Privacy Data Transformation API
De-identify your first piece of text in under five minutes.
1. Create an account and subscribe
Go to the Dashboard and sign up. Choose a plan — the Free plan includes 50,000 character credits, enough to get started.
2. Get your API key
After subscribing, your API key is generated automatically. Find it on the API Keys page in the dashboard. It looks like this:
cai_xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
3. Make your first request
Send a POST request to the de-identification endpoint with your API key in the X-API-Key header:
curl -X POST https://api.custodianai.com/api/v1/deidentify/text \
-H "X-API-Key: cai_your_key_here" \
-H "Content-Type: application/json" \
-d '{
"text": "Please contact Jane Smith at jane.smith@acme.com or call (415) 555-0123.",
"compliance_mode": "MASKED"
}'
Response:
{
"output_text": "Please contact **** **** at ******************* or call **************.",
"original_text": "Please contact Jane Smith at jane.smith@acme.com or call (415) 555-0123.",
"mode": "MASKED",
"meta": {}
}
4. Try Guardian Layer
Switch compliance_mode to PROPRIETARY to use Guardian Layer — CustodianAI's domain-aware detection that goes beyond standard PII:
curl -X POST https://api.custodianai.com/api/v1/deidentify/text \
-H "X-API-Key: cai_your_key_here" \
-H "Content-Type: application/json" \
-d '{
"text": "The Meridian account is managed by Sarah Chen in the London office.",
"compliance_mode": "PROPRIETARY",
"domain": "General"
}'
5. Next steps
- Compliance Modes — full breakdown of
MASKED,GDPR,HIPAA,CUSTOM, andPROPRIETARY - Guardian Layer — detect domain-specific sensitive content with the Guardian Layer proprietary mode
- Supported Entities — every entity type the API can detect
Python SDK
Deploy an AI assistant and start chatting in minutes.
1. Install the SDK
Requires Python 3.10 or higher.
pip install custodian-labs
2. Set your API key
export CUSTODIAN_SDK_API_KEY=cai_your_key_here
3. Deploy an assistant
from custodian_labs import create_assistant
app = create_assistant(
model="gpt-4o",
prompt="You are a helpful customer support assistant for Acme Corp.",
)
4. Chat
response = app.chat("What is your return policy?")
print(response.response)
The session ID is retained automatically — subsequent calls to app.chat() continue the same conversation thread.
5. Next steps
- Installation — env vars, constructor options, full setup
- Core Concepts — how assistants, apps, and sessions relate
- Assistant & App — multi-turn sessions, data sources, RAG