← Back to The Wire№ 4 · Sep 17, 2026
Shipped.
1 shipped this week

Every customer evaluating inline inspection eventually asks the same question: will this slow anything down? Issue № 4 — how we answered it without adding a single new probe.

Lead item

ZeroProbe™: experience monitoring, without the probes

π classifies every request it inspects — host, status code, and how long the whole round trip took, whether anyone ever asked for that number or not. ZeroProbe™ turns that side effect into an answer: opt a handful of apps in from π Config, picked from hosts π has already seen traffic to — not typed from memory — and π Inspection rolls up p50/p95 latency and error rate for exactly those apps. Hit Test nowand it re-runs on the spot. No bot pings anything on a schedule; no new traffic goes anywhere. The numbers only reflect real users, which also means a quiet app just shows no data — that's correct, not broken.

Turn on ZeroProbe™ →

Next up

→

π is moving from a single instance to a proper worker fleet, routed by nearest healthy node. That changes what a device needs to trust — a CA generated per-worker doesn't hold up once routing can land you on a different worker tomorrow. Making trust a fleet-level fact instead of a per-worker one is the next piece, not an afterthought.

The tempting version of this feature was synthetic checks — deploy bots, ping everything on a timer. We didn't build that. Every field a bot would measure, π already computes on real traffic, for free, as a byproduct of inspecting it. The feature was mostly deciding not to duplicate data we already had.— the Threatmatic team