Roblox enhances AI assistant with agentic tools for game development planning and testing

Roblox is set to enhance its Roblox Assistant AI tool with new agentic features, making it a collaborative ally for game developers. The upgraded Planning Mode evaluates game code and data models, poses clarifying questions, and converts prompts into actionable plans ready for execution. Two newly introduced AI tools, Mesh Generation and Procedural Model Generation, aim to streamline the development process. Mesh Generation enables creators to swiftly incorporate fully textured 3D objects instead of inferior placeholders, while Procedural Models allow developers to dynamically modify attributes of 3D models using code and the Assistant. Nick Tornow, senior vice president of Engineering, emphasized that these innovations bridge the gap between creative ideas and their realization. Planning Mode utilizes playtesting tools to analyze output logs, capture screenshots, verify design and gameplay, detect bugs, and offer automatic solutions. Roblox is also exploring the possibility of multiple AI agents working simultaneously, managing complex cloud workflows, and performing tasks like coding and testing. Furthermore, the company intends to integrate third-party tools such as Claude, Cursor, and Codex into Roblox Studio, establishing a self-correcting system that enhances accuracy over time through agentic loops.