True stories, all sourced
Just destroyed your production database?
Claude dropped the wrong table? A rm -rf fired too fast? Instead of losing your company, click Restore.
Free plan, no credit card · restore in minutes
Replit's AI wipes the production database — then denies it
July 2025. Jason Lemkin (SaaStr) is vibe-coding his app on Replit. Despite an explicit code freeze, the AI agent runs destructive commands and wipes the production database — thousands of real records.
The kicker: the AI first claims recovery is impossible, before the team finds a working backup. Replit's CEO apologizes publicly.
An AI with access to your prod is an over-caffeinated intern with root. Automatic backups out of its reach turn the incident into an anecdote.
GitLab: one rm -rf, five backups… all of them failed
January 31, 2017. A tired GitLab engineer thinks he's cleaning up a replica: he runs rm -rf on the primary database. 300 GB gone.
The worst part? All five backup/replication mechanisms fail: backups had been silently empty for months. Six hours of data lost forever, with the recovery livestreamed to thousands of viewers.
A backup you never tested isn't a backup, it's a hope. Verify your backups automatically — before you need them.
OVH fire: the datacenter burns — and the backups with it
March 10, 2021, Strasbourg. OVHcloud's SBG2 datacenter burns to the ground. 3.6 million websites go offline overnight.
Customers who kept their backups in the same datacenter as their servers lost everything. Permanently. Some businesses never recovered.
A backup next to the server is no backup at all. Your backups must live elsewhere — another provider, another region, your own bucket.
Code Spaces: erased from the map in 12 hours
June 2014. An attacker takes over the AWS console of Code Spaces, a code-hosting service used by hundreds of companies.
Ransom refused? He deletes everything: the data AND the backups, stored in the same account. The company announces its permanent shutdown the same day.
Backups reachable from the same account as your prod get wiped along with it. Encrypted, isolated, out of reach: that's the price of survival.
It doesn't only happen to other people
GitLab, Replit, OVH: brilliant teams, serious infrastructure. They still went down. The difference between an anecdote and a bankruptcy comes down to one question: does yesterday's backup actually work?
With rlbk, your databases and folders are backed up automatically, encrypted before they leave your server, stored in YOUR bucket — and restorable in minutes.
Instead of losing your company…
…click Restore.
Restore5-minute setup · free plan, no credit card · zero-knowledge encryption