Inspiration

My daughter always wanted to build a game so we sat together and saw that game-based learning measurably lifts math scores and motivation. Classic “brain-training” titles like Brain Age proved that even 15 minutes a day can boost arithmetic by up to 50 percent in primary classrooms. My daughter came up with the idea of a friendly safari narrative kids from grade 1 and 2 would love and would be inclusive of all kids that age.

What it does

Turns single-digit arithmetic into a bite-size safari: solve sums to help a lion hop stones, a zebra clear bushes, a monkey climb for bananas, and a parrot deliver berries

How we built it

One-Prompt Build in Bolt: A prompt outlined the tech stack, gameplay loop, and accessibility checklist in under 250 lines. Bolt assembled the project end-to-end on first run. Tech: Tailwind utility classes; Zustand handled global game state; Framer Motion tweened the animal journeys; Howler.js supplied cross-platform audio cues. Gameplay & Pedagogy: 3-question micro-levels with difficulty adapting on streak speed; every 4th item revisits a recent error. Progress stays in localStorage. Lighthouse + manual color-blind filters verified contrast and non-color cues for inclusivity.

Challenges we ran into

Token Budget: Cramming specs, UX notes, and rules into one prompt meant ruthless editing of game features. SVG Asset Size: Keeping animal sprites under 100 lines each to satisfy a single prompt generation. Animation optimization: Solved with Framer Motion’s useReducedMotion and fewer concurrent springs.

Accomplishments that we're proud of

Working on and generating a tightly-scoped prompt that helped us wrangle a complete kid-friendly math adventure in a single (Bolt.new) shot—ready for grade 1 and 2 students.

What we learned

Prompt engineering equals project engineering. We used Bolt’s docs info on using concise, file-by-file instructions for one-shot builds; which helped us generate a fully functional project that kids can enjoy!

What's next for Safari Math Journey

Integrate an Adaptive AI Tutor: Use a lightweight on-device LLM to generate hints and mini-explanations tailored to each error as well as use adaptive algorithms to drive measurable math-score gains.

Built With

  • framer-motion
  • howler
  • lucide-react
  • shadcn/ui
  • zustand
Share this project:

Updates