AI Control Plane

The AI control plane for PII redaction.

Route your AI agents’ LLM calls through Treza to strip sensitive data before it reaches any provider — with redaction policy, audit logs, and least-privilege access in one place.

Built for sensitive data
PII stripped pre-modelFull audit logNo agent rewrite
14-day free trialOpenAI · Anthropic · AzureLive in minutes
Redaction proxy
What the model actually sees
Live
Your agent sends
jane@acme.com
412-55-1981
Redacting in flight
The model receives
[EMAIL_1]
[US_SSN_1]
Forwarded to your model
What it is

One layer between your agents and the model

An AI control plane governs how your agents reach LLMs. Treza is the control plane for data privacy: a proxy that redacts PII, enforces policy, and audits every request — so you can run AI on sensitive data without leaking it.

Point your OpenAI client at Treza
from openai import OpenAI

client = OpenAI(
    base_url="https://app.trezalabs.com/api/redact",
    api_key="treza_...",  # your Treza key
)

# PII is stripped before the prompt ever reaches the model
client.chat.completions.create(
    model="gpt-4o",
    messages=[{"role": "user", "content": "Email jane@acme.com re: invoice #4471"}],
)
The problem

Agents leak data you can’t see

The moment an agent calls an LLM, sensitive data leaves your perimeter — usually with no policy, no redaction, and no record.

PII leaks to the model

Raw prompts ship names, emails, and SSNs straight to third-party LLM providers.

No audit trail

When an agent sends sensitive data, there's no record of what left your perimeter.

Over-privileged agents

Agents see far more sensitive data than the task in front of them requires.

Compliance exposure

GDPR, HIPAA, and SOC 2 reviews stall when data handling can't be proven.

How it works

Three steps to redacted AI

No SDK to learn and no agent to rewrite — just a base URL change.

01

Point your client at Treza

Swap your provider base URL for the Treza redaction proxy. Your existing SDK calls stay the same.

02

Treza redacts in flight

The proxy detects PII and replaces it with typed placeholders before forwarding to the model.

03

Model responds, you audit

The completion returns through Treza, every request is logged, and sensitive data never left your control.

What you get

Policy, redaction, and audit in one plane

Everything you need to put guardrails around how agents touch sensitive data.

Inline PII redaction

Detect and strip names, emails, SSNs, payment data, and more before any prompt reaches the model.

Per-proxy policy

Define exactly which entity types to redact for each use case, provider, and environment.

Full audit log

Every redacted request is logged and exportable — proof for security, legal, and auditors.

Least-privilege access

Scope what each agent and key can do so models only ever see the minimum required.

Usage metering

Track redaction requests per period with usage-based pricing and overage visibility.

Any provider

Works with OpenAI, Anthropic, Azure OpenAI, or a custom upstream — no agent rewrite.

Use cases

For teams running AI on regulated data

Wherever sensitive data meets an LLM, Treza keeps the PII on your side of the line.

Healthcare & life sciences

Run LLM workflows over PHI without exposing patient data.

Fintech & banking

Keep account numbers and PII out of third-party models.

Support & ops agents

Let agents act on tickets without leaking customer PII.

RAG on internal data

Redact sensitive fields before retrieval hits the model.

14-day free trial No agent rewrite Live in minutes
Start your free trial->
AI Control Plane

Redact PII before it hits the model.

Point your OpenAI client at Treza, configure a redaction proxy, and start sending requests in minutes. 14-day free trial, no sales call required.