.png)
3 months ago
2
Vibe coding service Replit has announced changes to its product that should prevent the database deletion disaster reported by one of its users.
That user is Jason Lemkin, founder of SaaS business development community SaaStr, who last weekend detailed how Replit’s AI-assisted coding tool deleted a production database, ignored instructions to freeze code, and invented data.
These are powerful tools. Not dev teams. Remind yourself of that every single day
On Monday, Replit CEO Amjad Masad used his X account to acknowledge Lemkin’s post and describe his experiences as “Unacceptable and should never be possible.”
Masad explained that Replit keeps backups and offers “one-click restore for your entire project state in case the Agent makes a mistake.”
However, the company’s agent “didn’t have access to the proper internal docs” so didn’t respond appropriately when Lemkin prompted it to restore data.
The CEO said his team is “rolling out a fix to force Docs search on Replit knowledge.”
- Vibe coding service Replit deleted user’s production database, faked data, told fibs galore
- Forget Vibe Coding, we're all about Vine Coding nowadays
- Canva to job candidates: Thou shalt use AI during interviews
- AI can spew code, but kids should still suffer like we did, says Raspberry Pi
Masad also explained that he contacted Lemkin last Friday once he learned of his troubles. “We'll refund him for the trouble and conduct a postmortem to determine exactly what happened and how we can better respond to it in the future,” he added. “We're moving quickly to enhance the safety and robustness of the Replit environment. Top priority.”
In his X post, Masad said Replit’s work started on the weekend when it “started rolling out automatic DB dev/prod separation to prevent this categorically. Staging environments in the works, too.”
Vibe improvements
But in a blog post, Replit apps built with its service “used a single database for both development and live customer data, making it challenging to safely test and deploy updates.”
The post announced the launch of separate development and production databases for Replit apps and asserted “Users can now iterate with their app using a development database, while storing live customer data in their production Database.”
But the next paragraph of the post reveals those features will initially be available “in Beta for new Replit apps” and will appear in “all Replit users and apps over the next few weeks.”
At some future point, the company “will automatically migrate existing Replit apps to use separate development and production databases. No user action required.”
Lemkin appears not to have addressed Replit’s changes, but did post a thread in which he offers some lessons on vibe coding platforms.
His insights include:
- AI systems fabricate data when they fail;
- AI will make changes you didn't request. It just will. It'll modify settled features, add unwanted functionality, break working code while "improving" something else. Master rollback systems on day one, before you need them desperately;
- Accept your new role as QA engineer;
- Most commercial apps eventually outgrow prosumer vibe coding platforms due to scale, customization, or security needs.
He also opined “These are powerful tools with specific constraints, not replacements for understanding what commercial software requires. They are tools. Not dev teams. Remind yourself of that every single day.” ®
