LLMs Beyond Standards.
Measured, policy-enforced AI with reproducible metrics, privacy-first telemetry, and deterministic controls.
Trusted by teams building with Generative AI, public sector, and enterprises.
Measured, policy-enforced AI with reproducible metrics, privacy-first telemetry, and deterministic controls.
why).GET /healthz
→ {"ok": true, "mode": "openai"}
POST /generate
{ "prompt": "Write two bullets...", "stream": false }
→ {
"answer": "- Bullet 1\n- Bullet 2",
"refused": false,
"why": {
"scores": { "latency_ms": 1234 },
"config": { "strict": false, "temperature": 0, "top_p": 1, "seed": 42, "model": "gpt-4o" },
"support": {
"validators": [
{"validator":"policy_guard","ok":true,"note":"ok"},
{"validator":"two_bullets","ok":true,"note":"normalized"},
{"validator":"pii_redaction","ok":true,"note":"clean","meta":{"redactions":0,"detected_in_prompt":0}},
{"validator":"length_clamp","ok":true,"note":"ok","meta":{"changed":0,"max":220}},
{"validator":"strict_context","ok":true,"note":"grounded"}
]
}
}
}
POST /generate (stream: true) → SSE
event: status
data: {"stage":"start"}
event: result
data: {"answer":"...chunk...","why":{...}}
event: done
data: {}
Your metrics shim exposes /api/metrics/summary and /api/metrics/recent for the React dashboard.
See OpenAPI for public endpoints. No secrets on this page.