- 1Add the recordsAn IEP, an eval report, a stack of either — Word, PDF, or scans.
- 2Two AIs read each oneDirect identifiers get scrubbed; contextual ones get underlined.
- 3You make the callsJudge the underlines against the neighbor test, then save.
Word .docx comes back as Word, formatting intact — including comments and tracked-changes authors, which get scrubbed from the file itself. .pdf and scans come back as text.
Scans work when the record is typed — handwriting can't be read.
First use downloads both AIs once (about 650 MB — the contextual one is most of it) and keeps them on this laptop. A long document takes a few minutes; the page stays usable while it works.
They arrive as one .zip in your Downloads folder. Word files stay Word files; PDFs come back as text.
The whole stack shares one set of tags — [NAME 4] means the same student in every record — so an AI can be handed several at once and keep track of who's who.
The WHO-IS-WHO file in the zip turns those tags back into your students — keep it, and never send it to the AI.
Got an AI reply about these records? Put the real names back — any record, one box
Because the tags are shared across the stack, a reply about any record — or several at once — un-scrambles here. Paste it below and the [NAME 4]-style tags flip back to the real words, on this device. No tags in the reply? Nothing to fix.
The underlines are the job. Colored tags are already scrubbed — click one to put the real word back if it's wrong. Underlined words are the neighbor-test calls only you can make: a diagnosis, a church, a team, a family detail. Click to scrub; the category chips scrub a whole group at once.
📷 This record was read from a photo or scan. The reader can misread print and can't read handwriting — give names, numbers, and spellings an extra-careful look before you share it.
The neighbor test is the bar: could someone who knows the family recognize this student? The tool finds; you decide. You're the last set of eyes.
Next step: paste the de-identified record into your AI — and if its reply uses the tags, bring it back here
The AI's reply will use the same [NAME 1]-style tags anywhere it mentions what you scrubbed. Paste the reply below and the tags flip back to the real words. No tags in the reply? Then there's nothing to fix — skip this. Exact tags only, and it all happens on this device, as always.