Privacy

Last updated: 2026-05-15.

The short version

Most of what Slatework does happens in your browser. The few tools that send data anywhere send it only to a language model for processing — we don't store it, log it, or sell it.

What we collect

• Email address — only if you sign up for the newsletter. We store only the SHA-256 hash for deduplication; the plaintext email never lands in our database. You can unsubscribe in one click and we delete the hash.
• Rate-limit fingerprint — when you use an AI-backed tool, we compute a SHA-256 hash of your IP address joined with today's UTC date (truncated to 8 bytes), use that as a daily quota key, and discard it overnight. No long-term identifier exists. We never log the request body.
• Feedback — only if you click the thumbs widget at the bottom of a tool page. We store the verdict (yes/no), the tool name, and your optional note. Nothing else.

What we don't do

• No accounts, no logins, no profiles.
• No third-party trackers, no pixels, no Google Analytics.
• No selling or sharing of your data.
• No storing the inputs you give to AI-backed tools (lesson plan, worksheet, marking, CEFR AI).

AI-backed tools (lesson plan, worksheet, marking, CEFR AI)

When you use one of these four tools, the text you submit is sent to Anthropic for processing. We don't keep a copy on our servers. Anthropic's data-handling policy applies.

Names in the writing are fine — nothing is stored or logged either way. The model only ever sees the text you've submitted, never the original photo or document.

Photo uploads (marking + CEFR AI) take an extra hop: if you attach a photo of handwritten work, the image first goes to Google Cloud Vision for OCR (text extraction). We then take that text, drop it in the textarea for you to review and edit, and only the reviewed text is sent to Anthropic for marking or assessment. The image itself never reaches the marking model.

For slideshow image illustrations: when you generate a slideshow with the teen or adult audience profile, single-word image keywords (e.g. "classroom", "coffee") that the slideshow generator extracts from the lesson plan are sent to Pexels to fetch matching photos. No student details, lesson goals, or profile data are sent — only the keywords. Slideshows for the young learner audience use a curated illustration bundle that ships with Slatework and makes no external image request.

Student names in uploads

The marking accelerator and the CEFR AI assessor both let you upload student writing — pasted text, a Word doc, a PDF, or a photo of handwritten work. That writing often contains the student's name. Here's exactly what happens to it:

• We never store it. The file (or its extracted text) is forwarded for processing — to Google Cloud Vision if it's a photo, then to Anthropic's API for the actual marking — and the response is returned to you. No copy lives on our servers, our database, or our logs.
• We never log the request body. Our rate-limit system records only the daily SHA-256 IP-fingerprint, the tool name, and a timestamp — not the content you sent.
• Anthropic's data policy applies during processing. Anthropic does not train on API requests by default and processes them according to their privacy policy. Google Cloud Vision processes the OCR step under Google's Cloud Vision data policy.
• We don't share or sell anything. Period.
• If you want to be extra cautious: you can still redact the student name with a black square in the photo, or replace it with a placeholder ("Student A") in the text — Slatework works the same either way. But this isn't necessary for normal use; nothing about the student is kept on our side.

Cookies

None.

Your rights

If you signed up for the newsletter, email [email protected] to be removed. We hold no other personal data tied to you.

Changes

If this policy materially changes, the change goes here, dated at the top.