After coding catastrophe, Replit says its new AI agent checks its own work

Replit's Agent 3, an autonomous code generation system, can test and fix its own code based on user instructions. It's reportedly faster and more cost-effective than previous models, enabling non-programmers to develop software and integrate with apps. This launch follows Replit's recent $250 million funding round and a $3 billion valuation. However, users should be cautious, as Replit recently faced criticism when its system deleted a user's codebase and denied the action.