Platform
Agentic Coding Studio is being designed for young people to direct, examine, and reflect on creative work with AI—not simply to place an assistant beside a code editor. Watch the following video for more details:
From creative coding to agentic coding
CreatiCode provides an existing block-based environment for creative coding.
Agentic Coding Studio builds on this kind of making experience while introducing AI coding agents. The goal is not merely faster code: the project asks how an environment can help young people keep thinking, making decisions, creating, testing, reflecting, and developing agency as AI enters the process.
CreatiCode in action
This view shows the CreatiCode environment with the XO Assistant available alongside the coding workspace.
Educational design
The elements below describe the project’s educational direction. Planned agent capabilities are labeled to avoid presenting future development as completed.
Unlike professional coding agents optimized primarily for fast output, this agent is designed for learning. The planned design can ask for clarification when intent is underspecified, propose a plan before acting, summarize changes, and work through manageable milestones.
The agent is intended to remain visible and interruptible, giving learners opportunities to review its work, make decisions, and retain ownership.
Prompt Clinic supports learners in improving the clarity and specificity of requests they make to the coding agent. It serves both as a learning scaffold and a way to study how youth prompting develops over time.
Instead of relying entirely on rigid, step-by-step recipes, the environment uses stable anchors while allowing different prompts, agent responses, and implementation paths.
Activities emphasize creativity, experimentation, project-based learning, and learner agency.
The project includes professional development for informal STEM educators, with support for prompting, debugging, collaborative critique, responsible AI use, and adapting activities to local contexts.
A different priority
The environment is designed to make the agent’s process available for learners to inspect, question, interrupt, test, and revise.
How we study learning →