AI product

Planning software that learns from one teacher’s edits

“Units for your classroom, not every classroom.” A teacher enters a grade, a subject, an idea and, if she likes, her town, and Alf drafts a full project-based unit for any grade from Kindergarten to Grade 12. I founded it, designed it and built it by directing AI coding agents.

Finalist, MIT Solve: Accelerate STEAM Education, The 2025 GM Challenge. Free during early testing for teachers in the US, UK and Dubai. Watch the walkthrough or try it

Alf drafting a unit, frame by frame — Fig. 1 Alf writing a real unit: a three-week Grade 6 science unit, done in 23 seconds by Alf’s own timer. Plays at double speed.

The product

Everything it drafts

One form: grade, subject, weeks, lessons, a topic and her town. Alf suggests standards by matching her words, with no AI, and she decides which stay. This is everything it drafted from that form.

The whole unit, from driving question to materials — Fig. 2 The whole unit, top to bottom: driving question, standards, overview, entry event, three weeks of activities, critique and revision, final product, rubric and materials. Choose a part, or scroll inside the frame to read it.

How it learns

Draft, teach, learns

Every save is a new version, and old versions can never be changed. Alf labels each change, such as “Added a scaffold” or “Changed the grouping”, in the background so she never waits. Only four habits can ever shape a draft: a tighter scope, sentence starters in week one, examples from her town, and groups of three. And only if she turns on “Start my drafts this way”, which is off unless she does.

Fig. 3 Draft, teach, learns. Alf only learns from her edits when she says so, in blue.

Alf's editing and learning screens — Fig. 4 She edits in place. Every line becomes a field on the same page, and nothing moves.

Walkthrough

The full walkthrough

Fig. 5 The real app, narrated, with captions you can turn on. Also on the Alf site

Limits

What Alf will not do

Six promises to teachers, each enforced in the code, not only in a policy.

Alf will not	How the product enforces it
Collect student data	There is nowhere to put it: the database has no field for anything about a student. A class is only a name for a group of units.
Mistake its ideas for yours	Changes she keeps from Alf’s own suggestions are counted, but never become habits that shape her drafts.
Make you chat	There is no chat screen. Every suggestion is a proposed change, checked against rules before she sees it, and thrown away whole if any check fails.
Lose your work	A saved version can’t be changed; the database refuses it. Unsaved edits are backed up in her browser.
Hide its sources	Every standards set records its publisher, source, date and licence, and a test fails if any notice is missing.
Keep your units from you	Print, copy into Google Docs or Word, or remove a unit at any time.

Quality

How it is checked

Tests: 1,131 automated tests, all passing, as of October 2026
Evals: Run by hand before changes. A reliability eval found 15 of 24 lesson builds failing; after the fixes it found, 0 of 30 failed, October 2026
Privacy: No student accounts or names. Her name and email are never sent to the model. No analytics or advertising
Built on: Next.js on Vercel, Neon Postgres, and Google Gemini through the Vercel AI SDK

Origin

From framework to product

NuVuX
Active Learning Framework
Catalyst, Issue, Method, Engagement. Co-developed with Nakeia Medcalf and Amanda O’Keefe; delivered virtually to leaders in Cameroon and Nigeria.
2023
AI co-design tools
Prompt packs and exemplars that drafted project briefs and rubrics for educators, checked and corrected by people.
2023 to 2024
AL and Di
Two custom GPTs: AL turns the framework into project-based curricula; Di guides teams through design thinking.
2025
Alf Coach
The first Alf.
2026
Alf Studio
One short form to a full, editable unit. Finalist, MIT Solve.

Fig. 6 Alf takes its name from the Active Learning Framework, which I co-developed at NuVuX.

AI in course production: agents draft, experts decide