For Program Reviewers & District Evaluators

Our standards crosswalks are published in full. In the K–5 reports, every FULL and PARTIAL verdict has a ▶ Play a live item link beside it; in the grades 6–8 and 9–12 reports, most do: Writing Coach rows can never be sampled, and some grades 6–8 rows have no link yet. Each link opens one real, watermarked practice item from the skill whose evidence earns the verdict — a sample from that skill, not a hand-picked item. No account, nothing saved on your machine. Three to start, each taken from the Kansas K–5 report:

• We never ask for a child's full name — the name box asks for a nickname or first name. For cloud accounts and classes, the server's student record holds that name, grade, avatar and progress. A Writing Coach answer can contain any name a child types.
• The app children use contains zero analytics, trackers, or ad SDKs — its security policy lets the app's own requests go only to its own site and its database provider (Supabase), whose platform logs record each request's IP address, browser and approximate location. A few features use outside services, each listed in the evidence pack below.
• No child data leaves the device through the app unless a family uses cloud sync, a student joins a class with a class code, or a student uses Writing Coach (browser voice features are covered below). The app also checks once per launch whether Writing Coach is available; that check carries no child data. Practice works offline; Writing Coach needs a connection. The in-app privacy receipt lists what family cloud sync sends, and an automated test checks that list against the family and reading sync code.
• No AI-generated text in front of children — all child-visible text is authored under human direction and human-reviewed, and there is no chatbot. An AI model (Anthropic's Claude) is used in one place: the Writing Coach check (paid plans and school-sponsored classes, from kindergarten up; in grades 6–12 it also grades the short written math explanations some math skills ask for). It scores each answer and picks from our fixed set of pre-written coaching lines; children never see text the model wrote. Our server sends the answer (up to 700 characters; up to 2,500 for the grades 9–12 essays), the prompt and the rubric, with no account, nickname or device ID added. AcornSmarts does not store the answer. Anthropic, which scores it, deletes API data within 30 days under its published policy (longer only if its safety systems flag a request or the law requires).
• Writing Coach safety label — the same scoring call can label an answer that seems to share private details (an address, phone number, email, or a full name or school name the prompt did not ask for) or first-person signs of real-life harm, abuse, self-harm, or serious fear or sadness. For a private-details label, the child also sees our pre-written reminder to keep private things private; for distress, the model is told to pick only gentle coaching lines. No screen tells the child a label was set. The child's words, the prompt, the date and the label are kept on that device only (the newest three per label, with no expiry date) and shown only in that device's paid parent dashboard, to whoever passes the grown-ups gate. That gate is the family PIN if one is set; otherwise it is a multiplication question an older student could answer. On a shared device, anyone who passes it sees every profile's notes. The notes are never sent to us, never synced, and never shown on a teacher screen. There is no switch to turn the label off. Districts should review Writing Coach with their counsel before students use it; that review may call for parental-consent language.
• Enrollment happens three ways — a family adds a child from its own account; a student joins a class on a device with the class code, a first name, grade and avatar (no account); or a teacher adds first names to the class roster.
• Voice and camera — voice answers (spoken answers and say-it-aloud items) are off by default, but the switch is in Settings, which opens without the grown-ups gate. When on, they use the browser's built-in speech recognition, which in many browsers may send the audio to the browser maker (for example, Google). Reading Buddy runs only where the browser can recognize speech on the device. Read-aloud (on by default) uses the device's voices; some voices the app prefers can be online voices run by the browser maker, and a few celebration lines speak the child's nickname. The app never records or stores audio; there is no audio-recording code in the application. The camera is used only to read an Acorn Card's QR code; frames are not stored or sent.

Full statement: acornsmarts.com/privacy. The full District Privacy Evidence Pack is public — data map, COPPA/FERPA positions, state-law notes (K.S.A. 72-6312 et seq.; §161.096 RSMo), and a DPA template for review by district counsel.

Single Student — $59/year (or $5.99/month): one child, all six subjects, Pre-K–8. This is our Pre-K–8 ESA-listed license (the $99 High School license is listed too), and it is identical to retail — we do not run special program pricing, which keeps every most-favored-pricing clause trivially satisfied. A $99/year High School plan covers grades 9–12, and family and classroom tiers and a one-time lifetime license exist alongside it. Pre-K–8 math and reading are free for one child, with placement, cloud save and the weekly parent digest — no card required. Program families redeem a pre-paid access key in the app, except on the High School license, which has no key yet: after the marketplace order we email the family, they create a free family account (if they don't have one yet) and reply with the email they used, and we switch the High School year on by hand — it starts that day. Either way: no card, ever, and nothing renews on them.

• Runs in the browser on any tablet, Chromebook, phone, or computer — app.acornsmarts.com, no installs, no app store.
• Bilingual (English/Spanish) Pre-K–12 — content and interface; High School skill names and descriptions are in English.
• Accessibility: WCAG 2.1 AA-minded design; read-aloud for early readers; dyslexia-friendly options — statement.
• Vendor: Black Swan Media Solutions LLC (Kansas). Contact: info@acornsmarts.com — answered within one business day.

A claims page is only as good as the machinery that keeps it honest after publication. Ours is automated and runs whether or not anyone is watching:

• At every release: the standards reports are regenerated from the same source code that builds the app — the crosswalk is re-verified against the live question banks, every play-link is re-resolved, and an automated guard blocks the release if the skill counts or standards figures on our pages disagree with the app's source. The public pages and the archival reports are checked byte-for-byte identical.
• Every day, automatically: a monitor re-checks the live site — the reports and this page are up, the 1,200+ play-links in the K–5 reports are still well-formed and the app still carries the route that opens them, checkout genuinely works (a real payment session is created and voided), and every paid account is re-reconciled against its billing source of truth. Any failure emails the founder the same morning.
• What we don't cover is listed as plainly as what we do — every NOT COVERED verdict and every honest gap note is on the same public page as the FULLs.
• The data is auditable: the exact machine-readable crosswalk files the reports are generated from are available to evaluators on request — audit our claims programmatically.

If you find a claim that doesn't hold up, we want to know first: info@acornsmarts.com.