Same lesson. The kid's world.
A research-preview / roadmap surface. AEGIS v1.0 classifies what content is. The generative gaming layer is the v1.1 roadmap target: same learning objective, restructured into the theme, vocabulary, and characters a specific kid cares about, with every generated frame passing AEGIS before display. R1 runtime integration is on the roadmap; today's experience runs in the FrameBright app without dynamic re-theming.
Three sub-stories
Auto-theme. Restructure. Classified.
1. Auto-theme — same lesson, the kid's world (roadmap)
The roadmap target: a spelling game becomes a dinosaur dig, an ocean dive, or a wizard's spell. Vocabulary, character names, settings, and reward art are regenerated per session from the kid's stated and observed interests. The learning objective stays fixed; the surface shifts to where the kid wants to be. This surface is in research; AEGIS v1.0 ships the classification loop today.
2. Restructure — the mechanic bends to the player (roadmap)
The roadmap target: a kid who reads at level 1.4 gets shorter prompts and bigger reward loops. A kid who skips intros gets a cold open. The skeleton (objective, difficulty curve, session pacing) stays; prompt length, reward cadence, and surface tempo flex. Adaptive flex is in research.
3. Classified — generated, then verified
The classification loop is live today: AEGIS v1.0 scores every piece of content on a 6-axis dual-judge rubric (Gemini 3 Pro + Claude Haiku 4.5) before it reaches the kid. When the generative re-theming surface lands in v1.1, every generated frame, line of dialogue, and reward asset will pass AEGIS before delivery — generate, classify, then deliver, never the other way around.
How re-theming would work (research) The classification loop (live)