Build AI pipelines
for regulated data.
Visually build and publish model pipelines on regulated data.
- FreeStarter plan1 published pipeline, 1,000 runs a month
- $99/moPro plan25 pipelines, 50,000 runs included
- $0.002Per extra runTransparent, usage-based overage
- 14 daysFree Pro trialNo card required. Cancel anytime.
Built for teams shipping AI on sensitive data
If your prompts touch customer data, Treza keeps it protected without slowing your team down.
Healthcare & PHI
Build pipelines over clinical notes, intake forms, and records with a redaction node so raw PHI never reaches a model.
- Redaction guardrail node
- HIPAA-ready patterns
- Model never sees raw data
Fintech & PII
Wire statements, support tickets, and KYC docs through a pipeline that strips SSNs, account numbers, and card data before any model call.
- PII guardrail node
- Full run history
- SOC 2-friendly trail
AI agents & copilots
Publish a pipeline behind one endpoint your agents call, with guardrails that keep customer data from leaking even when prompts are generated on the fly.
- OpenAI-compatible endpoint
- Guardrails built in
- No data exfiltration
Support & data ops
Summarize and classify customer conversations in a pipeline that redacts names, emails, and addresses before the model runs.
- Redact before the model
- Keep the insights
- Secrets stay encrypted
Go live in three steps
Ship a production-ready AI endpoint in minutes, with sensitive data protected and every run tracked.
- Step 01
Build a pipeline
Visually connect open-source models, guardrails, and integrations into a pipeline. Pick a model per node and drop in redaction or PII guardrails where you need them. No glue code.
$ drag & drop · no code - Step 02
Publish behind one API
Every pipeline gets a versioned endpoint. Call it with the typed /invoke API or an OpenAI-compatible /chat/completions endpoint — point any OpenAI SDK at it, streaming included.
$ POST /api/pipelines/<id> - Step 03
Track every run
Every run is logged with per-node timing, tokens, and success rate. Version, roll back, and export run history any time for SOC 2, HIPAA, or your next audit.
$ GET /v1/runs
Why we built Treza
A quick look at the problem we set out to solve. Start your free trial.
Everything you need to run AI in production
The AI Control Plane your engineers, CISO, and compliance team will actually approve.
Visual pipeline builder
Wire open-source models, guardrails, and integrations into a pipeline on a drag-and-drop canvas. No glue code between the pieces.
Learn moreOne API, two ways to call it
Every pipeline gets a versioned endpoint. Use the typed /invoke API or an OpenAI-compatible /chat/completions endpoint. Point any OpenAI SDK at it, streaming included.
Learn moreAny open-source model
Pick a model per node: Together, OpenRouter, Fireworks, DeepSeek, and more via an OpenAI-compatible backend. Swap models per node, never locked to one provider.
Learn moreGuardrails built in
Drop redaction, PII guardrail, and conditional routing nodes into any pipeline so sensitive data is handled before it reaches a model or external system.
Learn moreVersioning & run history
Version and roll back any pipeline. Every run is logged with per-node timing, tokens, and success rate. Export the trail for SOC 2, HIPAA, or your next audit.
Learn moreProduction-ready
Automatic retries and fallback models, a KMS-encrypted secrets manager, and hardware-attested Treza Enclaves on Enterprise, with provable evidence the right code handled your data.
Learn moreGovernance for every pipeline run
Every pipeline run lands in a tamper-evident audit log: who ran it, which model, what data it touched, and where the output went.
How we deploy it
We forward-deploy an engineer into your org to stand up the control plane with you and wire it into the identity and data systems your team already runs.
- 01
We embed an engineer
A Treza forward-deployed engineer works alongside your team to stand up the control plane inside your own environment. Not a slide deck, a working deployment.
- 02
Wire up your stack
We connect your identity provider (SSO / SCIM), your models, agents, and data sources, so every run is attributed to a real person.
- 03
Go live with full governance
Every pipeline run lands in a tamper-evident audit log you can stream to your SIEM. Your security and compliance teams own it from day one.
More than a pipeline builder
The same confidential-compute primitives behind the AI Control Plane are available to teams who need attested enclaves, key custody, or a control plane of their own.
Drops into the stack you already run
Ship containers you already build, sign for the chains you already use, and call the models you already trust, without changing how your team works.
What teams
ship with Treza
Real teams running AI on sensitive data today, from healthcare and fintech to autonomous agents, data ops, and identity.
Finance copilots
Pipelines that read statements, invoices, and support tickets and call a model, with a guardrail node that strips account numbers, SSNs, and card data first.
Autonomous agents
Agents that generate prompts on the fly call one pipeline endpoint with guardrails built in. It can’t exfiltrate customer data, no matter what they send.
Clinical AI
Run models on clinical notes and intake forms in a pipeline. A redaction node strips PHI before the prompt reaches the provider.
Audit-ready AI
Every run logged with per-node timing and a request id. Version, roll back, and export a tamper-evident trail for SOC 2 or HIPAA.
Data pipelines
Summarize and classify customer data in a pipeline that redacts raw names, emails, and addresses before the model runs.
KYC workflows
Verify users and screen for risk in a pipeline while keeping identity data out of the model. ZK-KYC primitives are available on Treza Platform to prove a check passed without storing the underlying ID.
Common questions
What teams ask before publishing their first pipeline.
What is the AI Control Plane?
It’s where you visually build model pipelines and publish them behind one API. Wire open-source models, guardrails, and integrations together on a canvas, then call the pipeline with a typed endpoint or an OpenAI-compatible, streaming one. Redaction and PII guardrails are nodes you add wherever you need them.
Do I have to rewrite my code?
No. Every pipeline exposes an OpenAI-compatible /chat/completions endpoint, streaming included. Point any OpenAI SDK at it: change the base URL and API key and you’re done. Prefer a native integration? Call the typed /invoke endpoint, with JSON inputs and outputs derived from the graph.
Which models can I use?
Any open-source model, picked per node: Together, OpenRouter, Fireworks, DeepSeek, and more via an OpenAI-compatible backend. Swap models per node so you’re never locked to one provider. Your keys are stored encrypted and never exposed to callers.
How do guardrails work?
Redaction and PII guardrails are nodes you drop into a pipeline. Add them wherever sensitive data should be handled before it reaches a model or an external system, alongside conditional and filter routing. A proxy is just a one-node pipeline.
Can I see what each run did?
Yes. Every run is logged with per-node timing, tokens, and success rate. Version and roll back any pipeline, and export run history any time for SOC 2, HIPAA, or your own records. Enterprise adds hardware-attested Treza Enclaves.
How much does it cost?
Starter is free with 1 published pipeline and 1,000 runs/mo. Pro is $99/mo with 25 published pipelines and 50,000 pipeline runs included, then $0.002 per run, plus a 14-day free trial that needs no card. Cancel anytime.
Put AI to work on your sensitive data.
Ship AI on your sensitive data and start free with full Pro for 14 days, no card required.