Cardinal Square’s AI writing teacher leaves the essay to the student

Cardinal Square is starting its plan for AI-native schools with a writing class whose central rule is that the student still does the writing. The company says its AI teacher guides children through an essay, reads their drafts and helps them correct mistakes without supplying the words for them.

The initial product is an after-school program for grades three through eight, priced at $29 a month in the company’s launch announcement. Students spend thirty minutes a day reading and writing, working toward a finished essay each week. Parents receive a weekly progress report and meet with an academic advisor monthly.

That is a narrower offering than the school in the company’s headline. Cardinal Square describes the current program as unaccredited supplemental learning. Full, accredited schools are a future ambition, not what families are signing up for at this launch.

The distinction makes the product easier to assess. For now, the founders are trying to make a familiar after-school routine available at a lower price: regular practice, a person or system that notices where a child struggles, and enough structure to finish something. They compare the service with writing tutors and learning centers, although the launch provides no independent comparison of educational outcomes.

Watch Cardinal Square’s original launch video.

In the company’s example of a fifth grader using the program, the student chooses a subject, develops ideas, makes an outline and works through drafting and revision. The AI teacher is meant to remember recurring mistakes and adjust its guidance over time. Cardinal Square says the system leads the child toward making the correction rather than completing the assignment on the child’s behalf.

That leaves a substantial task for the software. Remembering a spelling error is different from knowing when a student needs a new explanation, more practice or help getting started. The founders describe an ambition to teach as effectively as a strong human tutor, but the announcement does not include testing that establishes that level of performance.

Co-founders Bryan Hong and Jai Pal have known each other since middle school. In their launch post, they describe growing up in Asian American communities where tutoring and after-school programs were commonplace, then tutoring students themselves. Hong is a repeat Y Combinator founder who left Berkeley; Pal worked on machine learning at Snowflake and left the University of Texas at Austin, according to the announcement.

The program also keeps adults in the routine. Weekly reports are intended to show parents what their child practiced and what comes next, while the monthly advisor meeting provides a scheduled conversation about progress. The launch describes these as parts of the offering, but does not disclose enrollment, retention or independently measured improvements in students’ writing.

Cardinal Square’s broader plan is to rebuild more of the educational programs families already pay for around an adaptive AI teacher. The first test is more specific: whether a child can stick with daily writing, work through revisions and produce an essay that remains their own. For the families trying the initial class, that weekly piece of writing is the result they can actually inspect.