Roblox Assistant adds multi-step planning mode and AI-driven playtesting to automate game development

Roblox is deploying agentic features to its Assistant tool that plan, build, and test games through multi-step workflows. The company told TechCrunch the updates address failures of single-step AI tools that often miss creator intent.

Planning Mode replaces one-shot prompts

The enhanced Planning Mode transforms Assistant into what Roblox describes as a collaborative partner that analyzes game code and data models, asks clarifying questions, and generates editable action plans before executing changes.

When a creator prompts "create a park mini game with a fountain and foliage where characters have to collect coins," Assistant asks about visual style (cartoony, realistic, fantasy) and asset sourcing (build from scratch, use Creator Store models, or mixed approach). Creators can modify the plan and add context before implementation begins.

New 3D generation tools

Roblox announced two AI tools for the planning phase:

Mesh Generation creates fully textured 3D objects directly in the game world, eliminating low-quality placeholders during early development. Creators can prompt for a campfire, add realistic lighting, and set the scene at night through natural language.

Procedural Models (coming soon) generates editable 3D models with code. According to Roblox, Assistant understands 3D space and physical relationships, letting creators place and scale objects based on other scene elements. Attributes like bookcase shelf count or staircase height adjust dynamically.

Automated playtesting and bug fixes

As Planning Mode executes, it uses playtesting tools to read output logs, capture screenshots, simulate keyboard and mouse inputs to check design and gameplay, identify bugs, and provide feedback to Assistant for automatic fixes.

"Assistant is better at using agentic loops to test different aspects of the game, surface suggested solutions, and then incorporate the results into future planning loops, creating a self-correcting system that becomes more accurate over time," Roblox stated.

Future integration plans

Roblox is working on parallel multi-agent systems, cloud-based long-running workflows for complex tasks, and third-party tool integration. The company says it wants creators to "seamlessly use Claude, Cursor, Codex, and other third-party tools with Roblox Studio."

"The launch of our agentic features in Roblox Studio reduces barriers between creative vision and execution," said Nick Tornow, Senior Vice President of Engineering. "Assistant works as a multi-step, collaborative development partner — accelerating the process of planning, building, and testing."

What this means

Roblox is applying agentic AI patterns to a constrained domain where they may actually deliver value. Unlike general-purpose coding assistants that often produce broken code, Roblox's approach—clarifying intent before execution, operating within a defined 3D engine, and automated testing loops—addresses specific failure modes. The key test will be whether Planning Mode's multi-step verification actually captures creator intent better than iterative prompting with existing tools. Integration with Claude and Cursor suggests Roblox recognizes its proprietary models won't match frontier capabilities, making the orchestration layer its actual product differentiation.