The eight paths

Eight identities · Twenty milestones each · 160 total

Each path is an identity a kid can claim by doing the work. Milestones are ordered roughly by difficulty — the first five are the opening moves, the last few separate the committed from the serious. Earn five in a path and claim the identity. Earn all twenty and you are Proven.

The AI Pilot

An AI Pilot · 20 milestones
Proven at 10 · Mastered at 20
The kids who command the machines will have lives unlike the kids who merely use them.
  • 01
    Wrote Your First Prompt
    Used AI to do something useful.
    Ahead
  • 02
    Model Shopper
    Evaluated the same task across 5 different AI models, wrote a comparison of strengths and weaknesses for each.
    Ahead
  • 03
    Hallucination Hunter
    Caught and documented a factual error in an AI output, with a sourced correction.
    Ahead
  • 04
    Scheduled a Recurring AI Chore
    Set up an AI-powered task that runs automatically on a schedule and completes a real chore for you — a weekly digest, a monthly reminder, a daily summary — and kept it running for at least 30 days.
    Ahead
  • 05
    Wrote Code with AI That Ran
    Wrote working code with AI assistance that runs and does something useful, and can explain it line-by-line.
    Ahead
  • 06
    Built a Working Game with AI
    Built a playable game using AI — code, art, sound, or design — that someone else played start to finish and said was fun.
    Ahead
  • 07
    AI Tutor
    Used AI to teach yourself a genuinely hard skill from zero (language, instrument, math domain) and passed a standardized proficiency test in it.
    Ahead
  • 08
    Built a Tool a Friend Used
    Built an AI-powered tool that a friend or family member actually used to solve a real problem for themselves.
    Ahead
  • 09
    Connected an AI to an API
    Built a tool that connects an AI to a live external API and does something useful with the combined result.
    Ahead
  • 10
    Voice Clone
    Built an AI that produces work in a distinctive voice or style, and passed a blind test where evaluators couldn't reliably tell AI output from the source's real work.
    Ahead
  • 11
    Made Your AI Avatar
    Created an AI-generated avatar of yourself — image, 3D model, or animated character — that looks recognizably like you and that you used somewhere real (profile picture, game, video, project).
    Ahead
  • 12
    Agent Builder
    Built an autonomous agent that completes a multi-step task without human intervention — booking, research, monitoring, scheduling.
    Ahead
  • 13
    Tool Integrator
    Built a system that chains 3+ AI tools together (e.g., vision + reasoning + action) to accomplish something none could alone.
    Ahead
  • 14
    RAG Master
    Built a working retrieval-augmented system on a real document corpus — a working "ask my library" or "ask my notes" tool.
    Ahead
  • 15
    Local Model
    Ran a meaningful AI model entirely on your own hardware without cloud APIs — proved it works with a documented demo.
    Ahead
  • 16
    Fine-Tuner
    Fine-tuned an open-source model on a custom dataset, measured performance gains over the base model, and documented methodology.
    Ahead
  • 17
    Trainer
    Trained a model from scratch on a meaningful task — even small. You built the pipeline end-to-end.
    Ahead
  • 18
    Eval Designer
    Created a rigorous evaluation framework for an AI task that others have adopted or cited.
    Ahead
  • 19
    The Contributor · Any one of four
    Shipped something the AI community recognized.
    • Shipped an AI-powered tool used by 1,000+ unaffiliated people — not friends, family, or paying customers
    • had meaningful code merged into a major open-source AI project (transformers, llama.cpp, langchain, or equivalent)
    • won or placed in a recognized AI hackathon judged by practitioners
    • earned a medal in a Kaggle competition or equivalent recognized ML competition
    Capstone
  • 20
    The Authority · Any one of three
    Produced knowledge others learn from.
    • Published original AI research accepted to a recognized venue (workshop, conference, peer-reviewed journal) or cited by other researchers
    • published a technical paper, arXiv preprint, or detailed technical writeup that others have cited or referenced
    • taught AI skills to 20+ people through a documented course, workshop, or video series with verified completion
    Capstone