The Half The Race Is For

On Tuesday, July 14, Anthropic announced Claude for Teachers, a free, standards-aligned version of Claude for verified K through 12 educators in the United States.¹ The product went live with a library of teaching skills co-developed with the Chan Zuckerberg Initiative's Learning Commons and refined at Prospect Schools in Brooklyn.¹ Two built-in skills at launch: draft a lesson plan from vetted materials aligned to state standards, and differentiate materials for students at different readiness levels.² Free to any verified US teacher who signs up before June 30, 2027.² Detroit Public Schools Community District has been named as an early pilot site, in a study of educator well-being and practice.¹

OpenAI's ChatGPT for Teachers launched last November.³ Google's Gemini and Microsoft's Copilot both ship K through 12 versions.³ Three of the largest AI companies in the world are now openly competing for the teacher's seat in the classroom. Chalkbeat called it a race to become the default; The 74 called it a race to influence America's classrooms.¹⁴

Read the coverage a second time and something quieter shows up. The three products differ in polish, in integrations, in the promises they make about student data. What they do not differ in is the half of the classroom they were built for.

The Same Half

Benjamin Riley, founder of the education-policy group Cognitive Resonance, was blunt about it. He tested Claude's pre-made lesson-plan skill and said it "doesn't look any different than anything I've seen from OpenAI or from Google."³ Dylan Kane, a middle school math teacher quoted in the same reporting, said the risk is not that these tools help skilled teachers a little (they do), it is that they may de-skill newer ones.³

Whatever you make of those critiques, they land on the same axis. All three tools speed the teacher's day up. Faster lesson planning. Faster differentiation. Faster feedback drafts. Faster emails home. This is the axis the venture money is on, because it is the axis a teacher will happily use. The person doing the work and the person the labor-saving benefits are the same person. That is a clean market.

It is also, if you draw the timeline of a school day out, the "before" and the "after." The lesson plan is what happens before the student walks in. The graded essay is what happens after the student walks out. Between those two points is the thing the whole apparatus is supposedly there to serve: the student, in the middle of the work, thinking. That middle is the half of the room nobody in this race is building for.

Why The Middle Is Empty

There is a boring, structural reason for the emptiness. Instrumenting the teacher is easy. She is one person, with one login, doing one job. Instrumenting the student's process is hard. The student is thirty people, in thirty browser tabs, in three different applications, at four different reading levels, across a single hour. The signal is spread out and quiet. The lesson plan sits in one document. The thinking sits nowhere in particular.

There is a market reason too. The teacher will use a tool that gives her back an evening. The student will not use, and will not pay for, a tool whose value is the quiet, boring record of what she did on a Tuesday between 10:14 and 10:47. The district might. The district might not. The teacher is a clean buyer. The student's process is not.

The result is the pattern you see this month. Three of the largest AI companies in the world, competing for teachers, and none of them building for the space where learning actually happens.

What The Other Half Would Look Like

The other half, if you built for it, would not compete with any of these tools. It would sit next to them. The teacher would plan her lesson with Claude, or Gemini, or ChatGPT. She would differentiate it with the same. Her class would run. The students would write. Something else, quietly, would record the grain of what happened in that hour. Not the finished paragraph. The path to it. The pause. The revision. The moment one of them left the tab, came back a minute later, and rewrote the topic sentence differently. The sentence that arrived, whole, from another window, and did not get changed after arriving.

This is the layer we are building at Koan. It is not a race Anthropic, Google, or OpenAI is in. That is not a criticism of the three of them. Each of their tools is a good tool for what it is. It is a description of the shape of the market. The upstream and the downstream of the classroom have three companies competing to serve them. The middle has almost no one.

What The Fall Will Show

By fall, some version of one of these teacher tools will be inside most American classrooms. Lesson plans will be faster. Differentiated worksheets will be quicker. Feedback drafts will land in inboxes on Sunday nights instead of Monday mornings. The teacher's half of the room will run better than it has run in years.

The student's half will run the way it always has. She will sit down. She will do the work, or something adjacent to the work. The teacher will see what she turned in. She will not, unless someone builds it, see what happened between the prompt and the paragraph.

That is the gap the race is not running toward. It is also the gap most likely to define what this era of school actually did to the children who lived through it.

If your school bought every teacher-facing AI tool that launched this year, what would you still not be able to see?