Prova vs Galileo
Galileo charts how your pipeline is doing. Prova gives a verdict on each run, enforces the bounds, and signs the record.
BOOK A CALLGalileo is an LLM observability and evaluation platform: hallucination and quality metrics, prompt experimentation, and pipeline dashboards. Prova is the control plane. It observes too, but it also returns a pass / flag / needs-human verdict per run, enforces policy and boundaries at the gateway, and signs every decision into a verifiable audit trail.
| Feature | Prova | Galileo |
|---|---|---|
| Pipeline dashboards + quality metrics | per-run rollups | Yes |
| Deterministic per-run verdict | Yes | No |
| Hallucination / groundedness | Yes | Yes |
| Prompt experimentation | No | Yes |
| Gateway enforcement (block before the call) | Yes | No |
| Ed25519-signed receipts, verifiable offline | Yes | No |
| Runtime autonomy boundaries | Yes | No |
| Auto-rollback on a signed regression | Yes | No |
| Cross-customer fleet benchmark | Yes | No |
| EU AI Act / FDA / SEC / HIPAA export | Yes | No |
Where Prova is different
A verdict, not a dashboard
Galileo shows trends across a pipeline. Prova answers the operational question, did this specific run work, with a deterministic pass / flag / needs-human verdict and the signals behind it.
Observe and enforce
Galileo monitors. Prova monitors and blocks: a policy or budget breach returns action=block at the gateway before the call runs, and the block is a signed receipt.
Fleet intelligence
Prova's privacy-preserving fleet benchmark tells you if a model regressed for everyone after a vendor update, a signal you cannot get from your own pipeline alone.
Use Galileo for pipeline observability and prompt experiments. Use Prova for per-run verdicts, gateway enforcement, signed receipts, and fleet-wide regression intelligence.