How three of them fit together
From labor market to first class
A new program starts with a question: does this region actually need it? It’s done when someone can find the class, sign up and pay. Three things I’ve built cover that path, each picking up where the last leaves off.
The three cards in this story
The path, one stage at a time
01: Find the need
Does this region need the program?
Start with the region, not the catalog. Wavelength does the labor-market research a college rarely has time for: demand for the work, the programs that already exist nearby, and whether local employers would hire from it.
02: Test the idea
Is it worth building, here, now?
A feasibility study turns a hunch into a decision: go, revise or stop, before any budget is committed. Sometimes the right answer is not to build at all.
Inside the college, Pathway Builder is adding a place for the same question: a workspace that treats a new program idea like an investment to investigate, with the evidence beside it.
03: Build it
What should it teach?
A go becomes curriculum and courses. Wavelength develops them with the college; Pathway Builder keeps the competencies they teach in one library, with an AI-assisted writer grounded in O*NET data and reviewed by staff.
04: Map the path
Where does it lead a learner?
A new program shouldn’t be a dead end. Pathway Builder maps it into routes a learner can see, from short-term training to a degree, with non-credit-to-credit crosswalks and articulation records that keep their evidence.
05: Offer it
Can someone sign up?
Then it has to reach people. Encore puts the class in a college-branded catalog, takes registration and payment, runs the sessions, rosters and attendance, and issues a certificate on completion that anyone can verify.
Encore is still in development and isn’t in use by a college yet.
Inside Wavelength, roughly
I’ll keep the recipe to myself, but here’s the shape of it. Wavelength’s own line is the short version: tools do the reading, people stay accountable.
- Public labor-market data
- Federal and state data: wages and projections from the BLS, skills from O*NET, Census demographics, program completions from IPEDS, and state priority lists. It’s imported into Wavelength’s own database, so a number can be traced to where it came from and when.
- An MCP server
- A Model Context Protocol server gives AI tools read-only lookups into that data: wages, projections, skills, employers, completions and more. The AI works from the data, not from memory.
- Research, then persona synthesis
- AI researchers check a program against labor demand, competition, learner and employer demand, finances, fit with the college and regulation. Then AI argues it from several seats at the table, from product and finance to marketing and operations, before anything is summarized.
- Human review and direction
- A person picks the questions, weighs the evidence and sends the work back when it’s off. Drafts stay drafts until a person approves them, and AI analysis is never treated as verified fact.
- A decision you can check
- What comes out is a go, revise or stop recommendation with its sources beside it, and, after a go, drafts of the curriculum and courses.
Most of this is built and still being proven on real runs, which is one more reason a person signs off on everything that leaves it.
Where each piece stands
- WavelengthStatus: Live service
Live: research and program development for community colleges.
- Pathway BuilderStatus: Live product
Live. The idea-vetting workspace is being added now.
- EncoreStatus: In development
In development, tested with a sample college. Not in use by a real one yet.
Weighing a program of your own?
I like this part of the work best: figuring out whether something is worth building before anyone builds it.