Pathway Builder

Non-credit and credit teams plan in separate spreadsheets and documents. Handoffs drift, course equivalencies go undocumented, and new program ideas are often judged on a single labor-market report and a hunch.

Community-college workforce, continuing-education and academic staff: program developers, deans, and curriculum and articulation teams.

I set the direction, wrote the requirements and decision rules, and built it with AI coding agents under a release process where production changes need my explicit approval.

1. 01

   Treat program ideas like investment opportunities

   The opportunity workspace now being built is designed so a program idea can start as a record with only a title. Duration, tuition and capacity move from hypothesis to range to validated as evidence arrives.

2. 02

   Confidence never grants a go

   AI output is labeled as analysis. In the design, a final go/no-go has to be made by a person and linked to at least one piece of evidence.

3. 03

   Make “not equivalent” a real answer

   Competency mapping between non-credit and credit courses records full, partial, or reviewed-and-not-equivalent, so gaps are documented instead of silently skipped.

4. 04

   Useful without an account

   A set of free public tools (wage lookup, career explorer, skills-gap snapshot) works without signing up.

• Program, course and catalog import from CSV or spreadsheet, reviewed before anything is committed
• A competency library with an AI-assisted writer grounded in O*NET data, reviewed by staff
• Non-credit-to-credit competency crosswalks and gap analysis
• Regional program gap analysis using BLS, O*NET and CIP–SOC data
• A visual pathway canvas with comments, reviews and decision packets
• Articulation records with evidence and decision history
• Read-only public pathway links and printable records
• Free public tools at pathwaybuilder.app/tools

• Each college’s data must stay separate from every other college’s, enforced in the database as well as the app.
• Federal labor-market sources (BLS, O*NET, the CIP–SOC crosswalk) have to be cited, not smoothed into false certainty.
• AI can draft; people decide. Competency drafts and go/no-go calls need human review.
• Public claims about the product must match what is actually built. There’s a written claims policy, and a test that checks it.