Completion rate
90%
STEADYWRK closes 90% of dispatched jobs successfully on a rolling 30-day window.
Ground truth: contractor outcome + payment disposition.
▸ Public Evals · v2.0.0
The evals we run in production — published.
STEADYWRK publishes eight operational evals live on a rolling 30-day window: 90% completion rate, ±9% not-to-exceed (NTE) variance, a <2hr quote turnaround, 340ms median and 890ms p95 dispatch latency, and a 3% human override rate.
Ground truth is contractor outcome + payment disposition. Every number on this page is served from a public, no-auth endpoint — /api/dispatch/analytics/evals — so the page and the JSON cannot disagree. No other dispatch platform publishes this.
Verify it yourself — no auth, no key200 · application/json
curl https://steadywrk.app/api/dispatch/analytics/evals?period=rolling_30d▸ Eval registry
Eight evals, one rolling 30-day window.
Each metric is defined, sourced, and dated below. Quote any single line — every statement stands on its own.
NTE variance
±9%
Final invoices land within ±9% of the not-to-exceed (NTE) figure STEADYWRK quotes at intake.
Ground truth: quoted NTE vs. settled invoice.
Quote turnaround
<2hr
STEADYWRK returns a not-to-exceed quote in under 2hr from work-order intake — a target, not a binding SLA.
Ground truth: intake timestamp vs. NTE-returned timestamp.
Dispatch latency (p50)
340ms
Median API latency from work-order accept to contractor notification is 340ms.
Ground truth: server-side request traces.
Dispatch latency (p95)
890ms
Tail (p95) latency for routing plus contractor outreach is 890ms.
Ground truth: server-side request traces.
Human override rate
3%
A human operator escalates or reverses 3% of agent decisions; everything below 70% confidence is routed to a person by design.
Ground truth: operator audit log.
Policy-violation catch rate
Tracking
Currently in tracking: gated by Zod schemas and a policy layer on every decision. STEADYWRK withholds a headline number until the sample size is large enough to publish honestly.
Ground truth: pending sufficient sample.
Cost per decision
Private
Kept private to preserve contractor-margin confidentiality. STEADYWRK publishes operational quality openly but does not expose unit economics that would leak partner pricing.
Ground truth: internal only.
▸ Live readout
Operational snapshot.
- Jobs dispatched (30d)
- Active production
- Active providers
- 10 active
- Data source
- live · build f2569b0
- Operating states
- Multi-State
- Headquarters
- Aqaba
▸ FAQ
Questions, answered.
- What evals does STEADYWRK publish?
- STEADYWRK publishes eight operational evals on a rolling 30-day window: completion rate (90%), NTE variance (±9%), quote turnaround (<2hr), dispatch latency p50 (340ms) and p95 (890ms), human override rate (3%), policy-violation catch rate (in tracking), and cost per decision (private). Ground truth is contractor outcome plus payment disposition.
- How can I verify STEADYWRK eval numbers myself?
- Every number on this page is served from a public, no-auth, machine-readable endpoint: GET https://steadywrk.app/api/dispatch/analytics/evals?period=rolling_30d returns JSON. The page and the endpoint read the same canonical metrics registry, so the UI and the JSON cannot disagree.
- What is the STEADYWRK dispatch latency?
- Median (p50) dispatch latency from work-order accept to contractor notification is 340ms; the p95 tail latency for routing plus contractor outreach is 890ms.
- How accurate are STEADYWRK quotes?
- Final invoices land within ±9% of the not-to-exceed figure STEADYWRK quotes at intake, and a quote is returned in under 2hr from work-order intake (a target, not a binding SLA).
- Does a human review agent decisions?
- Yes. A human operator escalates or reverses 3% of agent decisions, and every decision below 70% confidence is routed to a person by design.