Krystal Jazmin Martinez · Head of Content, Curriculum & Education · a working demonstration

Not a course to browse.
A tutor that teaches you on the work you are already doing.

Claude Learn is a personal AI tutor: point it at your real task, and it teaches you one thing at a time — tied to a named standard, graded against ground truth, and remembered in a ledger that is yours. This site is the demonstration — the thesis, the standards, the evidence they came from, the playbook that generates the lessons, and forty-eight lessons you can read end to end.

0
adjudicated receipts
0
practice families
0
sample lessons
0
learner audiences
0
crosscutting concepts
0
proficiency bands
The argument

Seven chapters, one claim: the craft can be encoded, and here it is.

The job description asks for someone who can define what “great” means precisely enough that a team hits it without them in the room, decide what to trust AI with and where human craft is non-negotiable, and encode that craft into AI-powered systems rather than guarding it. Each chapter answers a piece of that. Read them in order, or jump to what you want to interrogate.

01 / Chapter

The thesis

Everyone is building courses. Claude Learn is a tutor — it teaches you on the task in front of you, one standard at a time. The argument, and the pedagogy underneath it: don't change the learner, design the environment so they succeed as they are.

Read the thesis →
02 / Chapter

What “great” means

You cannot hold a bar you have not named. The AIHC learning standards define AI-collaboration capability precisely enough that a team can hit it without the author in the room — 8 practices, 6 core ideas, 11 crosscutting concepts, 11 anchors, across 4 proficiency bands.

Explore the standards →
03 / Chapter

Craft that scales

The content team gets world-class at exactly two artifacts — the exemplar and the rigor rubric — and the human is removed from the moment of generation. This is what “encode your craft into the system rather than guard it” looks like, built.

Open the playbook →
04 / Chapter

The evidence

The standards are not asserted from taste. They were derived from 1,702 adjudicated verbatim receipts of one expert practitioner's real working record — distilled into 52 practice families, 38 of them proven across two tool eras. Paucity is a fingerprint, not maturity.

Inspect the families →
05 / Chapter

How it was derived

The full chain, in the open: NGSS/CCSS hybrid method, recall-first finder fleets under a published spec, byte-verified quotes, human-adjudicated structure with licensed dissent, full-corpus adjudication with documented tie-breaks. Rigor you can audit.

Read the protocol →
06 / Chapter

The proof

Four fictional learners — a developer, an enterprise admin, a consumer, a member of the public — each taught a 12-lesson arc on their own real work. Forty-eight lessons, every one standard-tied, doc-set-grounded, and honestly instrumented. Read them end to end.

Meet the learners →
07 / Chapter

The JD, answered

Not “explore and infer.” A direct crosswalk: each thing the Head of Content, Curriculum & Education job description asks for, mapped to the exact artifact on this site that proves it.

See the crosswalk →
How to read this site

You are on the hiring committee. This is built for that.

Throughout, amber notes guide the exploration — what to notice, why an artifact is here, where a claim is grounded, and where the honest limits are. Nothing is hidden behind a summary. The standards cross-link to the lessons that teach them; the lessons cross-link back to the standards and to the practice families that evidence them. It is a web of receipts, because that is the argument.