Starting Point Studio
A systems-before-code learning studio that moves AI creative programming from prompt-to-code generation toward system expression, milestone evidence, preview observation, and learner-owned revision.
Overview
Starting Point Studio is an AI creative programming learning environment and Systems-Before-Code Learning Studio. It reframes AI programming for children from "enter a prompt and generate code" into a creative learning process where students express a system, define completion signals, observe preview evidence, and explain how the system grows.
The project is also named 元点实验室 in Chinese. The name points to the first visible point where a learner's open idea becomes a project system that can be discussed, tested, revised, and extended.
Systems Before Code / Prompt-to-System
The core educational framework is Systems Before Code / Prompt-to-System. Instead of asking students to produce a full project through one large prompt, the studio asks them to make the system understandable before asking AI to generate or modify work.
Students learn to describe what should exist, how parts relate, what counts as a small finished ability, what evidence should appear in the preview, and how the system can grow from one milestone to the next.
Goal-to-Milestone workflow
The Goal-to-Milestone workflow turns open creative ideas into a sequence of visible design decisions:
- Goal understanding
- System relationship canvas
- Candidate system abilities
- Current making path
- Verifiable milestones
This lowers the cognitive threshold between "I have an idea" and "I know the next small thing I can make." It gives children a path into programming creation without forcing them to understand the whole project at once.
Prompt Workbench and First Milestone Studio
Prompt Workbench and First Milestone Studio turn prompting into a constrained experimental instruction. Before asking AI to generate or revise a work, students must clarify the current small ability, completion signal, preview evidence, and boundaries of the experiment.
This design prevents one-shot full-project generation. AI output becomes a response to a small, testable system question rather than a replacement for the learner's design process.
Learner agency and editable AI
The studio treats AI suggestions as confirmable, editable, and rejectable. The learner remains the author of the project system.
The system understanding trail records student wording, narrowed instructions, goal completion signals, preview changes, student judgments, and transferable system patterns. Over time, students can see not only what they made, but how their understanding of the system became more precise.