Ship & Sleep / Cases
Adopted and operated.
Three illustrative scenarios, told the way we report real engagements — drawn from the failure modes we see in AI-built apps, not yet from named clients. As real cases come with permission to publish, they replace these. What the Health Check found, what changed, and what the months since have looked like. Clients stay anonymous — their numbers don't.
Every webhook accounted for.
Subscription commerce · built with Lovable · Standard plan since March 2026
The situation
A founder built a subscription box shop with Lovable — real customers, real recurring revenue. Renewals started failing silently: Stripe webhooks timed out, retries weren't configured, and each missed webhook was a customer who paid nothing and noticed nothing. The founder found out from a spreadsheet that didn't add up.
What the Health Check found
Webhook handlers without retry logic or dead-letter handling, no alerting on failed deliveries, and a database with daily backups that had never been restore-tested. Two critical findings, four moderate.
What changed
Stabilization was a fixed quote from the report: a webhook queue with retries and alerting, monitoring on the renewal path, and a first restore drill. Then the app moved into Managed Operations on Standard.
Since then
Zero missed webhooks in six months of operation, 99.98% measured uptime, one incident — a Stripe API deprecation caught by monitoring, fixed within the included hours, postmortem #12 in the client's report archive.
The backups existed. Restoring didn't.
B2B booking platform · built with Bolt + Supabase · Priority plan since May 2026
The situation
A two-person company ran client bookings for trade businesses on a Bolt-built platform — the calendar was the business. They asked for a Health Check "just to be safe" before signing their biggest customer.
What the Health Check found
Daily backups were configured and green in the dashboard — and unrestorable. The backup target was keyed to the wrong region; every restore attempt failed. The dashboard had been reporting success for eight months. One critical finding, three moderate, one low.
What changed
The backup path was rebuilt and then proved: a full recovery rehearsed end-to-end, documented with timings. Because downtime directly cancels appointments, they chose Priority with its monthly drill cadence.
Since then
Restore drill: 14 minutes to full recovery, tested monthly, timings in every report. The big customer signed — the drill log was part of the pitch.
Three findings before the final invoice.
SaaS analytics tool · agency-built · independent acceptance check, July 2026
The situation
A founder commissioned an analytics dashboard from an agency and couldn't judge what was delivered. Before paying the final invoice, they ordered a Health Check as an independent acceptance check.
What the check found
API keys exposed in the client bundle, a data-loss path on concurrent writes, and unvalidated webhook input — three critical findings, each with evidence attached. Delivered in four days, while the invoice was still open.
What changed
The report went to the agency as the acceptance basis. All three findings were fixed and re-verified before handover and payment — no dispute, no lawyers, just evidence. The founder paid the invoice with confidence instead of hope.
Since then
The app runs with its original agency under a maintenance agreement — Ship & Sleep's role ended with the acceptance. Not every engagement becomes a subscription, and that's the point: the report serves you, not our pipeline.
Your app could be the next quiet one.
Every engagement here started the same way: one flat-fee Health Check and a straight answer.
Request a Health Check — $249