Learner-directed coding
Learners set meaningful goals, communicate intent, and decide what the agent should do next.
Learner-directed agentic coding
Create with AI.
Keep thinking in charge.
Agentic Coding Studio investigates how young people can build games, stories, and apps with AI coding agents—while staying responsible for the goals, decisions, testing, and revision.
A UC Irvine Digital Learning Lab research project
A research-based learning environment for creating with AI.
Agentic Coding Studio explores how young people can work with AI coding agents while keeping human judgment in charge. Agentic Coding Studio is part of the UCI Digital Learning Lab’s broader work on Computing and AI for All.
The learning cycle
Young people direct the process, not just the prompt.
Creation becomes a cycle of intention, review, evidence, and revision.
- Choose a goal
- Make a clear request
- Review the agent’s plan
- Build or revise
- Run and test
- Give feedback
- Reflect
What learners practice
More than getting code to run.
Computational agency
They use computational tools intentionally, critically, and creatively—and evaluate the results.
Learning through creation
Games, stories, simulations, and apps give ideas a form that can be shared, tested, and improved.
Learning in partnership
Beginning with community learning partners.
Initial implementation and research will begin with two organizations that support young people’s learning and creative participation.
Why now
As code production changes, computational thinking still matters.
When AI agents take on more mechanical production, people need to be better at specifying intent, decomposing problems, setting constraints, evaluating outputs, testing behavior, and revising systems toward meaningful goals.
Agentic Coding Studio does not treat AI as an answer machine or a replacement for thinking. It studies how learners can work with AI while retaining responsibility for what they make and why.