AI Gateway

The AI gateway built for data protection.

Route every agent’s LLM calls through a single endpoint that redacts PII, enforces policy, and audits every request — across OpenAI, Anthropic, and Azure.

One layer for every model call
PII stripped pre-modelFull audit logNo agent rewrite
14-day free trialOpenAI · Anthropic · AzureLive in minutes
Treza gateway
One endpoint, every provider
Live
Your agents call once
app.trezalabs.com/api/redact
Policy · redaction · audit
Routed to any model
OpenAI
Anthropic
Azure
PII stripped before it leaves
What it is

One endpoint between your agents and every model

An AI gateway is the proxy your agents call instead of each LLM provider directly — a single place to route requests, enforce policy, control spend, and log everything. Treza is the AI gateway built for data protection: it redacts PII in flight and audits every request, so you can run AI on sensitive data without leaking it.

Point your OpenAI client at the Treza gateway
from openai import OpenAI

# One endpoint. Any provider. PII redacted in flight.
client = OpenAI(
    base_url="https://app.trezalabs.com/api/redact",
    api_key="treza_...",  # your Treza key
)

# Route to any model through the same gateway
client.chat.completions.create(
    model="gpt-4o",  # or claude-3-5-sonnet, azure/gpt-4o ...
    messages=[{"role": "user", "content": "Email jane@acme.com re: invoice #4471"}],
)
Why you need one

Direct provider calls don’t scale

When every service calls providers on its own, there’s no shared policy, no spend visibility, and no record of what left your perimeter.

Direct calls everywhere

Every service and agent calls providers directly, so there's no single place to set policy.

PII ships to the model

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

No spend or usage visibility

Token usage is scattered across providers with no shared view of cost or volume.

No audit trail

When something leaves your perimeter, there's no record of what was sent or by whom.

How it works

A gateway you adopt with a base URL

No SDK to learn and no agent to rewrite — route through Treza and the controls come with it.

01

Point your client at Treza

Swap your provider base URL for the Treza gateway. Your existing SDK calls stay exactly the same.

02

Treza routes & redacts

The gateway applies your policy, strips PII, and forwards the request to the model you targeted.

03

You observe & audit

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

What you get

Routing, redaction, and audit in one gateway

Everything you need to put guardrails around how agents reach the model.

One endpoint, any provider

Reach OpenAI, Anthropic, Azure OpenAI, or a custom upstream through a single base URL — no agent rewrite.

Inline PII redaction

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

Per-route policy

Define which entity types to redact and which rules apply per use case, provider, and environment.

Scoped keys & access

Give each agent and key least-privilege access so models only ever see the minimum required.

Usage metering

Track requests per period with usage-based pricing and clear overage visibility across providers.

Full audit log

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

The landscape

How Treza compares to other AI gateways

Most AI gateways optimize for routing, caching, and cost. Treza is the one built for data protection.

Kong AI Gateway

Best for: Teams already on Kong's API platform who want to add multi-LLM routing, credential protection, and token rate-limiting to existing traffic.

Cloudflare AI Gateway

Best for: A cloud proxy focused on caching, observability, and token usage logs at the edge.

LiteLLM

Best for: A popular open-source, self-hosted proxy for fallback routing, load balancing, and budgeting across providers.

Portkey

Best for: A production GenAI control plane with a large library of guardrails and an MCP gateway.

TrueFoundry

Best for: An agentic gateway for platform teams to set budget caps and security policies without changing agent code.

Treza

Best for: Teams running AI on sensitive data who need PII redacted in flight, per-route policy, and an exportable audit trail. Use it on its own, or in front of any gateway above when data protection is in play.

Summaries reflect each product’s own positioning and are provided for comparison.

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.

FAQ

AI gateway questions, answered

What is an AI gateway?

An AI gateway is a proxy that sits between your applications or agents and large language model providers. It gives you one endpoint to reach any model, plus a single place to enforce policy, manage access, control spend, and log every request — instead of calling each provider directly.

What's the difference between an AI gateway and an API gateway?

A traditional API gateway manages generic HTTP traffic. An AI gateway is purpose-built for LLM traffic: it routes across model providers and applies AI-specific controls like PII redaction, token-based rate limits, and prompt and response logging.

Is an AI gateway the same as an LLM proxy?

They're often used interchangeably. “LLM proxy” emphasizes the routing layer, while “AI gateway” usually implies the proxy plus governance — policy, access control, observability, and audit.

Does Treza's AI gateway redact PII?

Yes. Redaction is Treza's core difference. It detects and strips names, emails, SSNs, payment data, and more from prompts before they reach any provider, and logs every redacted request for audit.

Which model providers does Treza support?

Treza works in front of OpenAI, Anthropic, Azure OpenAI, or a custom upstream. You point your existing client at the Treza base URL — no agent rewrite required.

How do I get started?

Swap your provider base URL for Treza, add your key, and send a request. Most teams are live in minutes on a 14-day 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.