How we will know whether the Foundry pilot worked

Foundry is inspired by adaptive-learning models that report unusually fast academic growth. Our central pilot expectation is at least twice the academic growth per focused hour that students would ordinarily experience. That is a target to test, not a current Foundry result or a guarantee for an individual child.

We are also testing whether students can sustain 120 minutes of verified focus, reach a high mastery threshold before moving on and finish a demanding shared capstone. A pilot can succeed on some measures and fall short on others.

Students will complete a baseline skill check before the academic program begins and a comparable check after week eight. During sessions, the system will record focused learning time, accuracy, skills practised and skills later verified without the original support.

We will separate engagement from learning. Time on task and earned credits can show participation; they cannot by themselves prove mastery. The before-and-after measure is designed to answer what changed, not merely whether the child enjoyed the software.

A small eight-week cohort cannot establish a universal effect. Families differ, students begin at different points and there may be no perfectly matched control group. Any published summary should therefore include cohort size, attendance, the assessment used, the calculation, incomplete data and changes made during the pilot.

We will compare our result with a benchmark set that includes a full-time US school operator, an after-school randomized evaluation in India, school-based trials in the United States and a short adaptive-learning experiment in China. None is a direct match for a San Kamphaeng Saturday pilot. If Foundry falls short, that is useful evidence. The purpose is to learn what the model can actually do here.

Families will also see the capstone, a progress summary and evidence of how their child approached challenge. Confidence, teamwork and motivation are important, but we will describe observations as observations rather than pretending every human outcome fits a clean number.

The standard is simple: make ambitious expectations visible, measure what can be measured honestly and let the evidence decide what Foundry becomes next.

A persuasive chart can conceal as much as it reveals. We will not present an average without the number of students behind it, remove incomplete participants without explanation or compare eight weeks of focused practice with an undefined idea of ‘normal school.’ We will distinguish operator-reported benchmarks from independent research and show why a short China study, a full-year US trial and an after-school program in India cannot be treated as identical evidence.

The central calculation will be growth per verified academic hour, because the pilot runs once a week while some reference models run every school day. Even that comparison needs caution. Frequency, retention, subject mix, assessment choice and the novelty of a pilot can all affect results. The honest output is not a magical number. It is a clear record of what we expected, what happened and what the next version must prove more rigorously.