The Line At Grade Seven

Last Monday, in a suburb west of Houston, a district's assistant superintendent stood in front of a board work study meeting and did not, quite, either ban or approve the tools that have kept every other district up all summer. She drew a line. Grade seven.¹

Katy Independent School District, which serves about ninety thousand students across dozens of schools, released its artificial intelligence framework for the 2026-27 school year on July 20 through Dr. Sanée Bell, its Assistant Superintendent for Teaching and Learning. It is not, quite, a policy. It is a graduated one.¹²

Kindergarten through sixth grade may use only two curriculum-embedded tools, Amira Learning and Writable, and only under a teacher's direct supervision. Google Gemini, Microsoft Copilot, and every other open-ended generative chatbot are off the table entirely for those grades.²³ Seventh grade opens a little. District-approved tools become available under supervision; non-approved chatbots remain off. Grades eight through twelve get the fullest access the district has drawn: generative AI for specific educational purposes, with teacher authorization and a requirement to cite it.³

The line is at grade seven. What sits on either side of it is worth pausing on.

The Question A District Actually Answered

Almost every American district this summer has been asked, publicly, some version of the same question. Should students use AI. The answers have arranged themselves in a narrow band. Some districts said no. Some said yes with conditions. Some said we will decide later. Almost none of them said what Katy's framework quietly says. The question is when.

Whether a fifteen-year-old writing a lit essay may consult a chatbot is not the same question as whether a seven-year-old learning to hear the sounds in a word may. A binary yes-or-no policy answers both at once and, by doing so, answers neither well. Katy's framework refuses to. It is a policy about developmental readiness that happens to be about software.⁴

The mission line in the framework document is one sentence. AI is a tool, not a teacher.³ The line is easy to nod at. What is interesting is the admission underneath it, that different students at different stages meet a tool with different equipment. A seventh-grader can hold a chatbot at arm's length in a way a first-grader cannot. That is not a rule about the software. It is a rule about the child.

What The Youngest Students Are Doing

The reason to draw the line at grade seven, if you have watched a first-grader learn to read, is the shape of the hour. A child at six is doing something a chatbot can effortlessly finish for her and cannot, by definition, do instead of her. She is sitting with the sound she does not yet know how to spell. She is starting a sentence she does not know how to end. She is holding the almost-word in her mouth and, if she is lucky, no one interrupts her.

Amira Learning and Writable are also AI. But they are narrow. Amira listens to a child read aloud and gives targeted feedback on decoding, fluency, and comprehension. Writable coaches a young writer through revision loops. Neither is an oracle. Neither answers for the child. They coach the pause without shortcutting it. That is what makes them safe at grade two in a way an open-ended chatbot is not.²

The developmental hour is not something a district can go back and re-attempt in tenth grade. It is a room that closes.

What A Graduated Policy Reveals

A district's most useful lever this fall may not be adoption speed. It may be adoption sequence. Which grade first, which tool where, which room later. Katy's framework does not settle the debate about AI in schools. It does something more useful. It admits that the debate is different in different rooms.

The most honest thing about the framework is what it leaves for the teacher to decide. Grade seven under supervision means the teacher has to be watching. Grades eight through twelve with authorization means the teacher has to be granting it. There is no shortcut around the moment when an adult who knows the student has to decide whether this student, on this assignment, is ready.³

That moment is what makes a policy real. It is also what makes it hard.

Where The Trace Belongs

At Koan we spend our time on the record of what actually happens while a student is doing the work. In first grade that record is small and, right now, largely invisible. A child rarely narrates why she crossed something out. She rarely explains why she left one page and came back with a different word. The trace is scattered across breaths and marks a teacher of twenty-eight children cannot possibly catch in the moment.

The developmental hour Katy is trying to protect is exactly the hour whose trace is hardest to see. The framework carves it out by keeping open-ended chatbots away from it. The next question, which no framework can answer on its own, is what the adults in the room will do with the hour once it is theirs to see.

When your district picks a grade, what is the room you are protecting, and how will you know if it stayed open?