Michigan OEAA / OEE Technical Advisory Committee
The April 2026 meeting, as a corpus
The April 23–24, 2026 TAC meeting produced a one-page agenda, five presentation artifacts, and nine pages of summary minutes. This site turns that material into something a committee can actually use between meetings — and argues that the whole TAC cycle should work this way.
98 corpus pages, fully cross-linked
37 discrete recommendations, each with a status
52 deck pages read as images to recover findings
2 items explicitly promised for the fall
The two documents
A Fuller Record
A comprehensive summary of the April meeting, organised by issue rather than by clock. Built from the official minutes treated as a rough transcript, plus everything the minutes could not carry — the findings locked in chart images, the eight hidden slides, the two speaker notes, and the participation total the handout never prints.
Written as a deliberate contrast to the official minutes.
Toward an AI-Native TAC
A brief for Andrew Middlestead, Director of the Office of Educational Assessment and Accountability, on what changes if the committee’s work is captured, structured, and carried forward as a corpus rather than as a document.
Part I makes the case. Part II is the technical appendix for whoever would build it.
The corpus
The corpus index is the way in. A few useful entry points:
- Open recommendations — the register. What the committee advised, to whom, and where each item stands. This is the page a working meeting should be run against.
- The meeting — roster, sessions, and the thread running through them.
- Glossary — AGP, NGA, CT-PO, QWK, MTTC, and the rest.
Standing topics outlive any one meeting and are the unit of institutional memory: adequate growth targets, subscore reporting, MTTC form security, interim adoption.
Why bother
The official minutes are competent. They are also lossy in ways that are specific and measurable: every question and every recommendation is attributed to “TAC” rather than to any of the five members present; a correlation is reported as “between 2.7 and 2.8”; the substance of a 52-page simulation study reduces to a single sentence because it lives in pictures; and material that exists only in speaker notes — including the repeal of educator evaluations from state law — never surfaces at all.
None of that is a failure of drafting. It is what happens when a two-day technical conversation has to become a nine-page document three weeks later, written by one person, with no way to represent a figure.
The alternative is not a better document. It is a different artifact: a corpus that any participant — or any agent working on their behalf — can query, that carries advice forward with its status attached, and from which documents are generated rather than authored.
This site is that alternative, built from the material we actually have.
This corpus is built from the April 2026 summary minutes treated as a stand-in for a transcript, because no transcript exists. Where the minutes attribute a question or a recommendation to the committee as a body rather than to a person, that attribution is preserved — inferring who said what would fabricate a record the source cannot support. That gap is the clearest single argument for capturing the next meeting properly.