Replit released the first iOS update for its Vibe-coding app, Agent 4, on Friday, May 10, following a four-month review block [1].
The return of the app marks a significant resolution between Replit and Apple. The dispute highlighted the tension between generative AI development and Apple's strict App Store guidelines, which can hinder the rapid deployment of AI-driven coding tools.
Apple temporarily blocked the application in March 2026 [1]. The company required changes to ensure the app complied with existing store policies before it could be redistributed to users [4]. This block prevented the app from receiving any updates for four months [1].
Amjad Masad, the CEO of Replit, confirmed the resolution of the conflict. "We worked things out with Apple," Masad said [2]. He also said that the app received its first iOS update in four months [3].
The friction comes amid a broader surge of AI-integrated software attempting to enter the mobile ecosystem. In the first quarter of 2026, 235,800 new apps were submitted to the App Store [6]. This represents an 84% jump in new submissions over a single quarter [5].
Some developers have expressed frustration with how Apple manages these guidelines. One developer said that Apple should either stop enforcing rules in a "weird way" or update the guidelines to accommodate the specific use cases of Vibe-coding [4].
Replit's Agent 4 is designed to allow users to build software through natural language and "vibes" rather than traditional manual coding. The resolution allows Replit to resume its update cycle and maintain its presence on the global iOS platform [1].
“"We worked things out with Apple."”
The conflict between Replit and Apple underscores the regulatory lag facing App Store reviewers as generative AI evolves. While Apple maintains a closed ecosystem to ensure security and quality, the 84% surge in new app submissions suggests a flood of AI tools that may not fit into legacy guidelines. Replit's ability to 'work things out' indicates a potential shift or a case-by-case negotiation process for AI agents that bypass traditional software development patterns.



