Metrics
32 L12 entries · across 3 chains
Time range
p50 latency
198 ms
p95 latency
372 ms
p99 latency
412 ms
Mean latency
188 ms
Latency distribution
32 samples · ms
6
0–52
0
52–103
4
103–155
8
155–206
5
206–258
4
258–309
3
309–361
2
361–412
Decision outcomes
Schema-valid
26/32
81.3%
Reviewed (human-in-loop)
14/32
43.8%
Fallback triggered
6/32
18.8%
Per-chain breakdown
| Product | Calls | Avg latency | Schema-valid | Reviewed | Fallbacks |
|---|---|---|---|---|---|
paysafe PSD3 + DORA | 12 | 165 ms | 83.3% | 2 | 2 |
cliniclens HIPAA + MDR + AI Act | 10 | 264 ms | 80.0% | 10 | 2 |
civicgate eIDAS + GDPR Art. 22 | 10 | 138 ms | 80.0% | 2 | 2 |
About these metrics
All numbers computed from the actual L12 chain files. No sampling, no aggregation pipeline — direct read of every entry. The substrate's substrate:watch-runtime-ai daemon emits the same metrics as JSONL on stdout for ingest into Loki / Splunk / Datadog.