Every August, in every school

You get eighty hours
and a subject
you’ve never taught.

Case Study · Danbury Public Schools

Building district-owned curricular experiences for 21 career pathways

Danbury Public Schools set out to give 30 high school teachers a way to write 21 senior-year pathway courses without losing the summer to it, and without the quality falling apart by Unit 6. The problem underneath is one every district running academies or pathways knows. This is the record of how Danbury built its way out, what the teachers said, and the first measured results.

★ Living document. The first result is measured and in hand (below); the per-teacher summer figures are captured as the 30 teachers write through the summer.

First measured result · observed June 29

90 minunder 15 minto review and give feedback on a full unit

In a live working session, the curriculum administrator reviewed a teacher’s completed unit, standards, Portrait, language-function tagging and all, in under fifteen minutes. By hand that same review runs an hour and a half. The teacher had built the unit in about two hours.

1

person built it

<4 wks

to build, custom

<1 hr

to onboard a teacher

faster unit review

30

high school teachers

21

career pathways

7

academies

100%

Danbury-owned

The problem

Every summer, Danbury hands teachers a course to write from scratch, often in a field they have never worked in: cybersecurity, clean energy, medical technology. The instruction is blunt and the math does not work.

We said, okay teachers, you have 80 hours, please write this curriculum course from start to finish. We all come back and you get a mishmash of a thousand different things, and they all look different.
Melissa Nadeau-Weinheimer, Ed.D, STEM Curriculum Administrator

Bringing those drafts into one coherent shape falls to her, by hand, across every pathway in the building.

That becomes the most ridiculous part of my job. Because it is impossible to do that work, just based on the sheer number of things we have in the hopper.
Melissa Nadeau-Weinheimer, Ed.D

And the work degrades on a predictable curve. The paid hours run out before the course is done, so the last units of the year arrive rushed and thin, frequently handed off to whichever teacher will actually deliver them. The standards, Portrait-of-a-Graduate, and multilingual-learner tagging, the part teachers dread most, is exactly where the quality slips first.

The intervention

The tool, Trellis, was scoped from a single instruction Melissa gave, and it is the reason the build is not a generic AI generator.

Take these five champions, find the commonalities in how they work and how their brain operates to create curriculum, and then make a tool that helps the other 25 catch up to them.
Melissa Nadeau-Weinheimer, Ed.D

The engine is the codified mental models of Danbury’s strongest curriculum writers, plus the industry and policy context the other teachers lack. The other 25 teachers start from the champions’ patterns as their floor. Around that engine, four design rules came straight from the teachers:

  • ·Carry the busywork in the background. The teacher writes in flow; the standards, Portrait, multilingual, language-function and assessment-type tagging happen underneath.
  • ·Write first, tag last. Teachers abandoned tag-as-you-go because it broke their flow, so the tool never forces inline tagging while drafting.
  • ·Suggest, then confirm. Every suggestion comes with a confidence signal and hover-provenance showing where it came from. The teacher always says yay or nay.
  • ·One surface, no tab sprawl. Standards and resources dock beside the writing field instead of living in a dozen open tabs and PDFs.

The line the teachers themselves drew: this is not a tool that hands you the answer. In Justin’s words, the differentiator is that it still tests your teachers. It removes the friction without removing the judgment.

The whole tool was built by one person, custom to Danbury’s own templates and champions, in under four weeks. And when the teachers first sat down with it, they were comfortable inside an hour.

What the teachers said

The build was tested in-hand, in working sessions, with teachers across the skill spectrum: a domain expert who dreads the standards bureaucracy, a heavy independent AI user, and a teacher with a rigid exam-driven pathway. These are their words.

All that other stuff is done in the background. It is a godsend.
Brian, culinary teacherOn Trellis carrying the standards, Portrait, and language-function tagging while he writes.
You nailed it. We feel like we have ADHD when we have to open tabs. It is actually a lot easier. It keeps you focused, blinders on.
Brian, culinary teacherOn the single-flow redesign replacing the old tab-and-PDF sprawl.
It is definitely set the way I plan. I really like this split now, that is cool.
Juliet, sports and human performance teacherUsing the unit editor live during a usability session.
This app has changed since last time I looked at it. This part excites me a lot.
Justin, academy leadA heavy independent AI user, on the post-redesign build.
If I can have something make a suggestion, and then for me to say yes, yay or nay. That takes away the load.
Elise, curriculum writerOn the suggest-then-confirm design that keeps the teacher in control.
For me to give feedback in less than fifteen minutes, that would normally take me at least an hour and a half per unit.
Melissa Nadeau-Weinheimer, Ed.DReviewing a teacher’s completed unit live, June 29. The first measured before-and-after.

The target being measured

The time savings is the headline; the codified champions are the engine. These are the four outcomes the summer’s real-usage data is being measured against. Until that data lands, the figures are stated as targets, not results.

OutcomeTodayTarget
Teacher time per course~80 hours~30 hours
Quality consistencyUnits 4-6 collapse late in the summerUnit 6 as strong as Unit 1
Curriculum review load~1,000 hours / year~200 hours / year
OwnershipRented or absentDanbury owns the code, data, and keys

Melissa’s own framing of what success feels like: to be released from typing into a curriculum document, to just review, edit, and create. That is my goal.

Why this is pathway curricular experiences

The match to the state-level mandate is not a framing choice. It is what the build literally produces. The 21 courses span seven career academies, from Scientific Innovation and Medicine to Clean Energy to Communication and Design. The artifact the tool generates is named, in Danbury’s own template, the Guaranteed Experiences Document.

  • ·Every unit carries a Career-Connected Lens and a Real-World Connection that ties the learning to the pathway’s industry, certifications, and careers. The pathway is not a label on top of a generic course; it is the spine.
  • ·Multilingual-learner equity is built in, not bolted on. Vocabulary cognates generate in both Spanish and Portuguese, and CELP language scaffolds are carried automatically, for a district that serves both populations heavily.
  • ·The work is backward-designed from a real end assessment, the way the strongest pathway teachers already build, so the experience leads to something students can show.

A district built this for itself, owns it outright, and can show measured outcomes, rather than waiting for a tool to be issued from above. That is a version of pathway curricular experiences other districts can build from.

The next build

This phase is scoped to the 21 senior-year high school pathways. The architecture does not stop there. The same engine, the same ownership model, and the same suggest-then-confirm surface extend in two directions, and both are decisions Danbury gets to make from a position of evidence rather than a cold pitch.

Deeper in Danbury

Extend from the senior pathways to full K-12 curriculum, bringing the same parity and time savings to middle and elementary writing. The hardest part, codifying how Danbury builds, is already done.

Beyond Danbury

A replicable model other Connecticut districts can stand up on their own templates and own outright, with Danbury’s build as the reference. The case study is the first instance of a pattern, not a one-off.

The open question for this study to help answer: what does the next scope look like, and what would it need to prove for the state? That is the conversation this document is meant to start.

Documentation timeline

Build (June)

Single-flow unit editor shipped and tested in-hand with culinary, sports/human-performance, and academy-lead teachers across the skill spectrum. Champion patterns codified into the suggestion engine.

Acceptance (June)

The tool moved into hands-on use by teachers and the curriculum administrator. Acceptance triggered by operational use, per the working agreement.

Summer (July to August)

Real-usage data accumulates as the 30 teachers write the 21 pathway courses for 2026-27. Per-teacher hours, draft variance, and last-unit quality are tracked against the targets above.

End of summer review

A consolidated review at the end of summer, with the measured before-and-after in hand.

Approval and privacy

  • ·All external publishing requires prior written Danbury approval, per the working agreement.
  • ·No student data appears in this study. Evidence is drawn from teacher and administrator working sessions only.

Busywork doesn’t change lives. You do.

CustosPracticeAI

Questions or feedback? brayan@custospractice.com