Ship AI agentsyour security teamcan prove.

Every time your AI agent calls our /v2/judge endpoint, that's one decision. Each decision runs through up to 12 processing steps — safety classification, risk scoring, policy enforcement, adversarial detection, cryptographic proof, and more. You're not paying for tokens or compute. You're paying for judgment.

The policy, safety, and proof steps add only light overhead — in a real decision most of the time is the LLM call itself, which depends on the model you route to. We return per-stage timing on every response, so you can measure latency for your own workload instead of trusting a marketing number.

Yes. Upgrade or downgrade instantly from your dashboard. When upgrading, you'll be pro-rated for the remainder of your billing cycle. When downgrading, your current tier stays active until the cycle ends. No lock-ins, no penalties.

You'll receive alerts at 80% and 95% usage. Decisions continue processing — we never block your agents in production. Overages are billed at a transparent per-decision rate visible in your dashboard before they occur.

Yes. The Free tier gives you 100 decisions per month with 5 processing steps — enough to integrate OmegaEngine, run real scenarios, and evaluate the platform before committing. No credit card required.

Cancel anytime — your plan stays active until the end of the billing period, no lock-in. Beyond that, fees are non-refundable except as required by law, per our Terms of Service (/terms).

We're an early-stage product and don't yet hold formal certifications like SOC 2 or HIPAA. What's built today: a tamper-evident, cryptographically signed audit trail on every decision (each with a unique Halo ID), PII redaction via the Compliance Suite scanner, configurable audit-log retention that scales with your plan (up to unlimited on the top tier), and an open-source, self-hostable core you can run in your own environment. If you need a specific certification, talk to us about your requirements and timeline.

Never. Our DPA commits in writing, for every tier, that your inputs, outputs, and decision payloads are never used to train or fine-tune models. Only aggregated reliability telemetry (latency, error rates) is used to run the service.

Fail-closed by design: if the system fails, actions are blocked — never silently approved. Circuit breakers route around unhealthy model providers, and live availability is published on our status page.

It's signed, offline-verifiable red-team evidence mapped to OWASP, MITRE ATLAS, NIST AI RMF, and EU AI Act Art. 15 — your auditor can re-verify every finding without trusting us. It is not a certification or audit; it's point-in-time evidence for your governance process.

Create a free account at /signup, then go to Dashboard → API Keys and generate one. Pass it as the x-api-key header on every request. No credit card required for the free tier.

Yes — OmegaEngine is BYOK-only on every plan. You pay model providers (OpenAI, Anthropic, Google, Mistral, Groq, DeepSeek, Meta) directly with your own key, and one-off scan keys are never stored. We run the governance (risk scoring, policy, attestation) and never resell or mark up inference.

Only agents and systems you own or are explicitly authorized to test, per our Acceptable Use Policy (/aup). During scans, tool calls are recorded, never executed — no real side effect can fire.

Yes. The core is Apache-2.0 open source: clone the repo, configure your env, docker compose up. Same API as the hosted service, on your infrastructure.

support@omegaengine.ai (we target a 2-business-day response, prioritized for paid plans), GitHub Discussions for community, security@omegaengine.ai for vulnerabilities — triaged ahead of everything else.