Inspiration

Year 9 girls are picking subjects that quietly decide their careers — and most have no idea what Methods, Specialist, or Physics actually unlock. Counsellors are stretched thin, parents often can't help, and cultural pressure pushes girls away from STEM before they've had a real look. We wanted a product that felt like an older sister explaining what these subjects are actually like, told through the voices of women who've already walked the path.

What it does

STEMpath is a web app that helps middle and high school girls choose VCE STEM subjects with confidence. Students browse subjects as colourful cards and open a detail page with five AI-summarised sections — overview, what it feels like, why people recommend it, skills built, and where it leads — synthesised from real reviews by uni students and women working in STEM. A career recommender turns a VCE subject plan into matched STEM career paths, a public forum lets advisors answer student questions, and "Ask Pip," our AI guide, translates vague feelings ("I like art and physics scares me") into concrete next steps.

How we built it

We started with a design doc covering voice, palette, and every screen's inputs and outputs, then built the front end in React on Replit with a pastel system tied to subject identities. AI summaries and Pip's responses run through the Anthropic API with prompts tightly scoped to advisor data — so the model rephrases real reviews instead of inventing facts. Subject cards, careers, and reviews live in a structured data layer; AI calls are cached per subject and refreshed when new reviews land.

Challenges we ran into

Getting the AI to sound warm and direct — not like a brochure — took several prompt rewrites and a long banned-words list ("amazing," "girl power," "you've got this"). Designing for honesty without being discouraging was harder than expected: we wanted to say "Methods is hard" without scaring anyone off. Balancing safety on the forum (no DMs, named advisor roles only, distress-detection in Pip) against a chill, friendly vibe also took real care.

Accomplishments that we're proud of

A product that genuinely sounds like it was made for teenage girls, not at them. A subject-card system where colour, tags, and difficulty all carry information without feeling clinical. AI prompts that refuse to invent advisors or sugarcoat hard subjects. And a clear path from "I don't know what I like" to "here are three careers and the subjects to get there" — in under a minute.

What we learned

The biggest unlock isn't more information — it's permission. Girls need to hear other girls say the hard parts out loud before they'll consider a subject. We also learned how much voice matters: the same advice in a textbook tone gets ignored, but in a friend's tone it lands. And technically: tightly-scoped AI prompts with curated source data beat clever prompts on open knowledge every time.

What's next for STEMpath

A "subject swap simulator" that shows in real time which careers close or open as you toggle subjects in your plan. Weekly role-model spotlights so representation is pushed, not just searched for. A "what your parents might say" companion piece for each subject, tackling cultural pressure head-on. Class composition data ("~28% of Specialist students are female — here's what 4 of them said") to shrink the fear of being the only girl. And expansion beyond VCE to NSW, QLD, and WA curricula.

Built With

  • apis
  • replit
Share this project:

Updates