Every teacher deserves to know who understood.

India has ten million teachers. Retraining at scale is a decades-long effort. But a teacher with real-time misconception data outperforms one without it. The tool does the diagnostic work that normally takes years of pedagogical training — surfacing what to reteach, and why, in one sentence.

Re-targeting instruction to what students actually know is among the best-evidenced interventions in development economics: Pratham’s teaching-at-the-right-level trials (Banerjee et al., 2017), a “great buy” in the 2023 GEEAP Smart Buys review, and Mindspark’s 0.37 SD in Delhi (Muralidharan, Singh & Ganimian, 2019). Formative assessment more broadly: a meta-analysis finds 0.2 SD, less in mathematics (Kingston & Nash, 2011); the older narrative review claimed 0.4–0.7 SD (Black & Wiliam, 1998). This tool delivers the diagnosis half of that mechanism in a 60-student classroom with one smartphone and a blackboard.

A student’s first page is under 100 KB with fonts, about 10 KB a page after; it loads in seconds on 2G. Offline-capable — answers queue in the browser and sync when signal returns. No app install, no login, no password, no name. Six languages. Students answer on any phone, or on none: the teacher counts raised fingers (live today) or scans printed answer cards (in development). The design constraint was Chhattisgarh, not Bangalore.

The school dashboard aggregates misconception patterns across all teachers, subjects, and grades. District and state benchmarking creates an accountability loop between classroom practice and system-level outcomes — a loop that does not exist today.

Classroom Sims is live and no classroom has used it. There is no monitoring data, no adoption data and no learning-gains data of our own. The district report in the pilot film uses demonstration figures.

Diagnostic reports and low-stakes monitoring were randomized across 200 rural Andhra Pradesh schools: teachers worked harder when observed, and after a year students still did no better on independent tests (Muralidharan & Sundararaman, 2010). That feedback came once, at the student level, with no action attached. This tool differs on all three: every lesson, at the class level, with a reteach suggestion. Whether that difference produces use is the first thing a pilot must measure.

Do teachers use it without incentives? Do they act on it? Which delivery model produces use at what cost? What would a school-randomized trial need? A pilot is not powered to show impact and will not claim it. The evidence page sets this out.

Embed into state education department workflows and DIET teacher orientations. Zero marginal cost per school.

Train block resource coordinators and cluster resource persons to introduce the tool during school visits.

A first pilot in two districts that measures whether teachers use it and act on it, what it costs to run, and what a randomized trial would need. We claim no adoption data today.

Add Kannada, Malayalam, Gujarati, Odia, Assamese to reach every medium of instruction in India.