Observed arrival · 2026-10-25
Trace Margin asks whether the voice agent actually finished the job
Independent research and practical tools help US buyers evaluate AI voice agents by completed work, cost, and handoffs—not just how convincing they sound.
○Open
⊠Login
$Paid
†Ads
✦Pretty
●Pro
◎Niche
◉Human
⚑Risk
ƒJS
- For
- US operations, finance, and procurement teams
- Worth noticing
- Its sample plumbing call is explicitly synthetic, and the page warns that a booking request is not the same as a confirmed system write.
Field notes
The evaluation path moves from listening to a sample call through workflow selection and cost modeling, then into research and buyer tools. Queue examples are framed with operational boundaries, including exclusions, destination systems, and a named fallback. The provider list also marks some entries as unverified, making uncertainty visible rather than presenting every listing as equally established.
Observed signals
Read the marks
Editorial observations of this landing page, not a rating.
○OpenPublic substance visible
✦PrettyNotable craft visible
●ProPolished or operationally mature
◎NicheUnusually specific use
One card from the complete issue