JobShadow AI

Try any career in 20 minutes before you commit years to it. AI-generated real job scenarios, 4-option decisions, scored by how you actually think; not what you studied.

Comment

What inspired us

Every year, millions of people make one of the biggest decisions of their lives — what career to pursue — based almost entirely on guesswork. They read job descriptions, talk to people at careers fairs, maybe do a personality quiz. Then they spend 3-4 years and tens of thousands of euros training for a role they have never actually experienced.

Internships don't solve this. You only get an internship after you've already committed to a field. We wanted to build something that exists before that decision — a way to feel what a job is actually like before you bet your life on it.

The research backs this up. A 2025 peer-reviewed study (CareerPooler, arXiv) proved that AI-driven career simulation significantly improves engagement, information gain, and career clarity compared to traditional tools. The academic proof existed. The consumer product didn't. So we built it.

What it does

JobShadow AI puts you inside the cognitive experience of any career in 20 minutes. You pick a role, either VC Analyst, A&E Doctor, Criminal Lawyer, Product Manager, Architect , get briefed with real context, then face 7 genuine workplace decisions with 4 options each.

The decisions are designed carefully:

  • Two options reveal different thinking traits (analytical vs empathetic, cautious vs bold)
  • One option is high-risk, high-reward
  • One option is the expert trap — subtly wrong in a way only someone with real domain instinct would catch

At the end, you receive a personalised Career Fit Report scoring your risk appetite, empathy, analytical depth, and domain instinct. No quiz. No personality type. Just how you actually performed.

How we built it

The entire product was built in 24 hours at the Future Innovators Hackathon in Lisboa. Stack:

  • Next.js 14 — frontend and API routes
  • Claude API — the AI scenario engine generating branching, stateful career simulations
  • Supabase — session management and results storage
  • Cloudflare Workers — edge caching for job templates and API routing
  • Tailwind CSS — UI
  • jsPDF — personalised career fit PDF generation on completion

The core of the product is a single well-crafted system prompt that instructs Claude to generate realistic, high-stakes workplace scenarios for any given career, maintain state across 7 decisions, and produce a structured scoring report at the end. Getting this prompt right — ensuring decisions felt genuinely difficult, the expert trap was subtle enough to fool overconfident users but not careful thinkers — was the most critical engineering challenge of the build.

Built With

Share this project:

Updates