Platform reliability & monitoring

WallaB.AI monitors its own production health so problems are found and acted on before they reach you. This page describes, honestly, what that monitoring actually is.

#Durable event logging

Every meaningful operation — auth flows, subscription changes, billing attempts, AI concierge calls, webhook processing, and all server errors — writes a durable, structured event with a request-correlation id. Your own shop's events are visible to you on the Activity page of your admin dashboard; our operations team watches the same event stream fleet-wide.

#Background jobs retry themselves

Billing runs, dunning retries, webhook processing, and campaign sends all run through a persistent job queue. A failed job retries automatically with exponential backoff; only after repeated failures is it marked permanently failed — and permanently failed jobs are surfaced directly to our operations team for a human decision. Money-path jobs are idempotent, so a retry can never double-charge a shopper.

#Automated operator alerting

A server-side rule set continuously evaluates fleet health — new permanent job failures, worker crash loops, database connectivity errors, error-rate bursts, and email hard-bounce spikes — and emails the platform operators the moment a rule trips. Each rule has a cooldown window so a sustained incident produces a steady, readable signal instead of an email storm.

#Email deliverability

Every outgoing email's outcome (sent, failed, bounced, complained, or suppressed) is recorded. Hard bounces and spam complaints automatically add the recipient to your shop's suppression list, and deliverability is monitored per shop so one sender's problems can't degrade everyone else's reputation — see Email domains for how sending domains are isolated.