Skip to main content
Skip to the work
SteplatchSynthetic checks for form, checkout, and booking paths — with step evidence.

Journey monitoring

Catch the broken customer step before it stays invisible.

Give Steplatch an ordered form, checkout, or booking path with expected and actual observations. It stops at the first mismatch, opens an incident only after the same step fails twice, and records recovery only from later passing runs.

Sign in and you go straight back to the Steplatch conversation. Work happens in chat. This page never asks for your card details or email address.

FIRST RESULTA synthetic checkout or booking journey run with every step result, the first failure evidence, and separate recovery proof when resolved

01 / AUTHOR THE PATH

Bring an ordered path, not a live card.

Teams that need forms, checkout, and booking journeys tested as real step sequences with recovery evidence.

  • 01

    An ordered journey with expected observation per step

  • 02

    Actual observations from a synthetic checkout or booking run

  • 03

    Recent run history and, after a repair, later passing runs

02 / THREE COMMANDS

Uptime scores hide the first broken customer step. A later page can look fine after an earlier timeout, and a verbal “it works now” is not recovery.

  1. 01diagnose

    diagnose walks steps in order and stops at the first actual/expected mismatch.

  2. 02decide

    decide opens an incident only after two consecutive failures at the same step.

  3. 03recover

    recover keeps passing run ids in input order and requires two passes before recovered.

WHAT THE BOARD RETURNS

Recovery is a later stamp, not a verbal all-clear.

Return 01

status=failed plus the first mismatched step, or status=passed

Return 02

open_incident, close_incident, or observe from consecutive runs

Return 03

verified passing run ids, recovered_at, and recovered or not_proven

CONTROLLED PACKAGE EVAL

Three synthetic cases were scored against the delivered script.

Catalog evidence from eval-tasks/critical-path-monitor. These are not production booking or payment outcomes. environment=synthetic.

diagnose
failed_step=create_paymentexpected=test_payment_created · actual=timeout
decide
action=open_incidenttwo consecutive submit_form failures
recover
status=recoveredverified_run_ids RUN-8, RUN-9 · recovered_at 2026-07-14T10:20:00Z
Incident
INC-9 · checkout-demoverified_run_ids keep input order

YOU STILL HOLD THE KEYS

Steplatch scores the observations you supply. It does not charge a real card.

For teams that need forms, checkout, and booking journeys tested as real step sequences with recovery evidence.

It does not

  • Live payment charges or real customer credentials
  • Treating a single acknowledgement as proof the path recovered

Before acting

  • Synthetic runs must use approved test identities and avoid real payment data.
  • A passing check proves only the defined path and region at that time.

BEFORE YOU AUTHOR A PATH

Questions

What do I need to start?

An ordered path with expected results and the actual observations from a synthetic run. The engine scores those records; it does not charge a real card.

When does it open an incident?

Only after two consecutive failures at the same step. A single timeout stays observe so a flaky run is not treated as an outage.

When is recovery proven?

After at least two later passing runs. The recovery record keeps those run ids and the last passing finish time. Acknowledgement alone is not recovery.

STARTER / MONTHLY

Starter

Billing appears inside the signed-in workspace. Card payment runs through Stripe Checkout.

US$19 per month

This landing page never captures card or email details.

Continue to workspace billing
Steplatch

Signing in opens the Steplatch conversation. This page uses PostHog for product analytics (anonymous, optional). See Privacy.