Potato Class is a K-12 math and reading platform for classroom practice, a digital reading library, math manipulatives, formative class tests, and a research-informed adaptive placement assessment called Grow. It runs in the browser at potatoclass.com, desktop or tablet — no install. A playful character layer — a customizable potato avatar, a voice assistant named AiPa, six educational mini-games — keeps K–8 students engaged without getting in the way of the work.

Reading is built on the Simple View of Reading (Gough & Tunmer, 1986), operationalized through Scarborough’s Rope (2001) and the National Reading Panel’s five pillars (NRP, 2000). Math covers the five strands of mathematical proficiency (NRC, 2001) in a concrete–representational–abstract sequence (Witzel, Mercer, Miller, 2003). Content is aligned to the Virginia SOL with crosswalks to 11 state standards frameworks that together cover all 50 states + DC: Common Core (~41 states), plus VA SOL, TX TEKS, FL B.E.S.T., GA (2023), IN, OK, NE, MN, SC, and AK. 100+ peer-reviewed citations at growingstandard.com/research.

A per-domain Rasch-Elo computer-adaptive engine with standard-error-based stopping rules and calibration-aware item selection (50% target student ability, 30% explore the near-boundary, 20% sample uniformly to preserve calibratability) — the same family of measurement architecture as iReady and NWEA MAP Growth (Lord & Novick, 1968; Embretson & Reise, 2000; van der Linden & Glas, 2010). Empirical IRT calibration (Rasch model first, with 2PL escalation only if a pre-registered likelihood-ratio test justifies the added parameter) follows the spring 2028 district pilot, when sample size supports defensible parameter estimation; the routing engine is designed today so the pilot data is calibratable when it arrives. Pre-pilot validation rests on two pre-registration analysis-plan drafts deposited at OSF — a concurrent-validity study (Grow placement vs. state SOL, teacher judgment, iReady / MAP / STAR) and a practice- engagement-and-growth correlation study (does in-game engagement predict learning growth, decomposed across six telemetry vectors). Formal Registrations (the locked, DOI-minted snapshots) are targeted for July / September 2026 after academic methodologist review and before any pilot data flows. Grow is untimed, matching iReady and MAP Growth. Three windows per school year (Fall, Winter, Spring). Results include a per-domain ability estimate (theta) updated after each non-rapid-guess response, with a confidence interval from the per-item Rasch-Elo standard error; per-skill proficiency; growth bands (High / Typical / Low); and reading-level equivalents in Fountas & Pinnell, Lexile, DRA, and Grade Equivalent scales. Rush detection flags responses below grade-adjusted speed thresholds per Wise & Kong (2005), regardless of whether the answer was correct.

We think districts deserve to know where a new assessment tool actually is in its evidence base, not where it claims to be. So:

• Today. Item parameters are theoretically calibrated from tier-based difficulty estimates, not empirically calibrated from student data. Our sample size is pre-cohort.
• Primary validation signals at this stage. Two pre-registration analysis-plan drafts deposited on OSF, with formal Registrations (the locked, DOI-minted snapshots) targeted for July / September 2026 after academic methodologist review and before any pilot data flows. Study 1 is concurrent validity against external benchmarks — teacher judgment, iReady, NWEA MAP, state SOL — with pre-committed correlation thresholds. Study 2 is a practice-engagement-and-growth correlation study asking whether in-game engagement predicts learning growth, decomposed across six telemetry vectors so the contribution of each (practice, library reading, math games, lessons, class tests, grow sessions) can be evaluated separately. Either study clearing its thresholds supports a Tier 3 (Promising) ESSA classification; both clearing strengthens the package.
• What we commit to post-pilot. Empirical recalibration via mirt::multipleGroup(), Mantel-Haenszel DIF analysis, conditional SEM reporting, classification accuracy, and a transparent validity writeup.
• Tier 2 horizon. ESSA Tier 2 (Moderate) requires a quasi-experimental design with a comparison group and is not reachable from correlational evidence alone. The practice-engagement-and-growth study is the natural input to a future Tier 2 effort with partner districts, realistically 2028 onward. We will not claim Tier 2 evidence until a comparison-group design is run and published.

Full doctrine at growingstandard.com/research and in our published measurement-science appendix.

Growing Standard LLC is founded and led by Daniel Segal, a public-school classroom teacher in Fairfax County, Virginia. One person today. Self-funded. The question pipeline runs on teacher-authored specifications, not runtime AI. No outside capital and no pressure to oversell. If you want a direct conversation with the person who designed the question pipeline and built the engine, that person is me, and my inbox is open at daniel@growingstandard.com.