Cheeklay

Every district promises its community goals, and buys programs and software to reach them. Most of it never reaches the classroom: across 48 districts, a typical app had 30% of its licenses never used even once. The teachers who have to carry the goal are rarely asked: only 31% say they have a lot of say in how school decisions are made, and 93% say they should.

Three steps: start with a goal the district has already set money aside for, sit with the teachers who have to reach it, then build the fix with them, with AI that guides their work and never does their thinking. In Danbury that became Trellis: reviewing a unit went from 90 minutes to 15, and unit creation dropped from 80 hours to 45. Each tool we build joins a portfolio the next district can start from, while each district keeps its own work.

AI can now guide a teacher's work without doing their thinking, so a tool can be built around one district's own templates and review rules. And with federal pandemic relief money gone, districts are cutting what can't prove it gets used. If another product were the answer, districts would already be where edtech promised. Another product isn't needed. A different kind of company is.

Districts have been sold a lot that never got used, so they distrust a new vendor. Starting like a design firm, in the room with each district's teachers, is slow, and the portfolio only pays off once a second district reuses what the first one built. Danbury worked because I grew up there and sat with its teachers. The hard part now is the second district, and the tenth.