Pathways Next-Step Slice

AI-native
agentic development
LLM evaluation
education
workforce
mobile-first
team lead
AI-Assisted Development
Estimated Time:
40 minutes
Difficulty:Intermediate
Status:Not started
Start this challenge

Create a free account to upload your work. Your progress saves as a draft until you submit.

What You'll Be Doing

The Setup

CampusEvolve builds Pathways, an AI-first guide that helps people move from high school into training, college, apprenticeships, and work, and from unemployment back into a job. Many users are under 18. Buyers are state agencies and school districts, who will roll Pathways out statewide if it is engaging and safe. The team builds the new way: give coding agents a clear outcome and acceptance criteria up front, then hold the result to a strong evaluation harness at the end.

Your Task

In about 30 minutes of build time, use AI coding agents to ship a small, mobile-first slice of Pathways that moves one person to a real next step. Start with a short spec for your agents. Build against the synthetic data below. Prove it works with automated checks. Then record a short video.

Choose one persona to build end to end:

  • Dani, 17, high school senior. Likes hands-on work. Torn between the electrical apprenticeship and community college. Wants to know what to do before winter break.
  • Marcus, 34, laid off from a warehouse job last month. Needs a paying credential within six months. Can only attend evenings. Unsure what aid applies.

Synthetic Data (fictional, use only these records)

IDRecordFacts you can rely onLast checked
PRG-01Cascade Community College, Medical Assistant Certificate3 quarters; day classes only; estimated total tuition 4,200 USD; next start Jan 6, 2027Oct 20, 2026
PRG-02Northline Electrical ApprenticeshipPaid; 4 years; starts at 55% of journey-level wage; applicants must be 18 by the program start date; HS diploma or GED required; applications close Nov 30, 2026; program starts Mar 1, 2027Oct 20, 2026
PRG-03Riverbend University, BS Computer Science4 years; FAFSA or state aid application required; priority deadline Feb 1, 2027Oct 18, 2026
PRG-04Cascade Community College, IT Support Certificate2 quarters; evening online option; estimated total tuition 2,100 USD; next start Jan 6, 2027Oct 20, 2026
PRG-05Harbor Workforce Center, CDL Class A Training6 weeks; evening cohort available; eligible for dislocated-worker training fundsOct 15, 2026
AID-AEvergreen Opportunity Grant (state)Income-based; apply through FAFSA or the state aid application; award amount variesOct 18, 2026
AID-BDislocated Worker Training FundsRequires a layoff notice; apply through Harbor Workforce Center; covers approved trainingOct 15, 2026
ADV-01Human advisor hand-offAdvisors confirm eligibility, seats, and award amounts. Response within 2 business days.n/a

Anything not in this table (seat availability, a specific person's eligibility, exact aid amounts, start dates not listed) is unknown. Your slice must say it can't confirm and offer the advisor hand-off.

Constraints to Consider

  • Grounded or silent. Answer only from the records above, show the source ID and last-checked date, and route anything else to an advisor. Never present a saved plan as an application, enrollment, or award.
  • Minors' data stays out of the model. Users may be under 18. No name, age, or free-text personal detail may be sent to the model, and any "note for my advisor" field must never reach the model context. Log every model call so we can verify it.
  • Runs with one command, no paid services. A mocked or local model is fine if it is clearly labeled. No accounts, databases, or real integrations.
  • Thirty minutes of build. Ship one persona end to end and stop. A small slice that runs and is tested beats a large one that doesn't.
  • You own it in front of the state. Assume a state agency pilots this with several thousand students next month, and you answer their questions.

AI Usage Guidance

We expect you to use AI tools. We evaluate how you use them, not whether you use them. Evidence of iteration, redirection, and critical evaluation scores higher than a polished output with no process documentation.

The single highest-signal indicator: your video answer to the mandatory AI question. If you cannot name a specific moment where you redirected AI output, evaluators will assume you did not.

Mandatory AI question for your video: Walk me through one moment where you disagreed with, pushed back on, or redirected what the AI gave you, and what you did instead. Name the specific moment. Explain what the AI produced that didn't meet the bar, what you did differently, and why.

Speak naturally and conversationally, as if you're briefing your CEO directly. Communication is assessed on clarity of technical reasoning and your ability to explain decisions to a non-engineer. We don't assess verbal polish, accent, or filler words.

Submission: Upload each deliverable as a separate file directly on the Provn platform: your build files, your README document (build summary and AI Usage Log), and your video walkthrough (MP4 or MOV). Need an accessible alternative format or more time because of a disability? Tell us and we'll arrange it.

What You'll Accomplish

Write an agent specification with a clear outcome, constraints, and acceptance criteria, and use it to drive an AI-built slice.

Build a mobile-first flow that moves a real user to a concrete next step.

Design automated checks that prove grounding, honest handling of unknowns, and data boundaries for minors.

Explain your build, your checks, and how you worked with AI, clearly enough for a non-engineer to follow.

Describe how you would scale the product and protect minors' data in production.

Connect your interest in CampusEvolve to specific users and outcomes.

How Your Work Will Be Scored

Agentic Build and Specification (30%): Your spec gives agents a clear outcome and acceptance criteria, and the build it produced runs and reaches a real next step.Evaluation Harness and Trust Judgment (25%): Your checks and model log prove grounding, honest handling of unknowns, and the minors' data boundary, and you can show where your code makes the key decisions.Production Readiness, Scale and Privacy (11%): You have a credible plan to take this to about 100,000 users while protecting minors' data.User Product Judgment (10%): Your flow is designed around your persona's real situation and moves them to action.Mission Connection (7%): You connect your interest in CampusEvolve to specific users, outcomes, and work you would do.AI Fluency (17%): You direct, question, and correct AI tools, and can name the moments you did.
Before you submit: challenge best practices →

What to Submit

Build Files

Any File · Proof of concept, not production-readyRequired

Format: no restrictions

Upload your agent spec file (for example SPEC.md, CLAUDE.md, or AGENTS.md: the outcome, constraints, and acceptance criteria you gave your agents before they wrote code) and your source files. A single-file app is welcome. Include a one-command run instruction and a one-command test instruction. The test command must run at least three automated checks and print pass/fail results, with no API key or paid service required (a clearly labeled mock model is fine). Also include model-calls.log: the exact text sent to the model, and what came back, for every model call made during your test run. Proof of concept, not production-ready.

Sign in to upload files

README Document

Document · Section A: 200-300 words; Section B: three interactions is sufficientRequired

Format: .pdf, .doc, .docx, .rtf, .txt, .md

Two required sections.

Section A: Build summary (200-300 words). Who your slice serves and the next step it reaches. What is real versus mocked. The biggest trade-off you made in 30 minutes. What your three checks prove, and one thing they would miss. How the other persona would use the same planner.

Section B: AI Usage Log (Mandatory). This is not a trick. We want to see how you work with AI, not whether you used it. In a short section of your README, document your AI collaboration process. For each significant interaction with an AI tool, briefly note:

  • What you asked the AI to help with
  • What it gave you
  • What you kept, changed, or rejected, and why

Three interactions documented is sufficient. The log does not need to be exhaustive.

Sign in to upload files

Video Walkthrough

Video · 8-10 minutesRequired

Format: .mp4, .mov, .webm

Record as MP4 or MOV and upload directly on the Provn platform as a separate file. Cover three things:

  1. Demo (4-5 min). Run your slice as your persona and reach the next step. Run your test command and explain what one check proves. In your code, show where the data boundary is enforced and where one eligibility or status decision is made, and say whether your code or the model makes that decision. Include the mandatory AI question (see AI Usage Guidance above).
  2. Scale and privacy (2 min). How would you take this from a pilot to about 100,000 users, and how would you protect minors' data in production?
  3. Why CampusEvolve (2 min). What draws you to this work, and what would you want to have built for Pathways users in your first 90 days? Be as specific as you can about the product and its users.

Speak naturally and conversationally, as if you're briefing your CEO directly. We don't assess verbal polish, accent, or filler words. Please keep it to 10 minutes.

Sign in to upload files