59% more growth than projected, for $41 a student

Math is our weakest area, and it starts early

The area we think is weakest for us is mathematics, and the challenges start early. Our entering 6th graders often lack foundational skills, number sense, how numbers work, and they aren’t used to talking about mathematics or seeing the world through a mathematics lens.

So when we get into grade-level content, there’s a huge gap between grade-level expectations and a poor mastery of foundational skills. It’s like the equalizer bars on a stereo: which bar do you move up and down to balance the need for foundational skill and the need for grade-level content? That’s the balancing act our teachers face every day.

I test-drove it in the sandbox

We use an online curriculum, and this was actually the first year we started both a new math curriculum and Skye. When I heard Skye was an AI tutor, I wanted to test-drive it, so we got access to the sandbox. I was looking at the latency, the mathematics language, and the kinds of prompts it used.

The only students who grew below the norm did few sessions

After one semester with Skye, our 6th graders grew 17.2 RIT points from fall to spring against a projected 10.8, placing the cohort at the 66th percentile for growth. 69% of students met or exceeded their growth projection, against the 50% you’d expect.

And volume made the difference in our results. The only students who grew below the national norm were the ones who did a low number of sessions. Students who did more than 20 sessions grew well above projection.

Scheduling is where AI tutoring proved its worth

Alongside Skye in 6th grade, we also ran face-to-face tutoring with college students in 7th and 8th grade, and the logistics were a constant battle. We couldn’t start until February, college and school schedules never quite lined up, and we simply couldn’t get the hours in.

By comparison, Skye is far easier to schedule, and with high-dosage tutoring, getting the scheduling right is usually the hardest part. We ran two sessions a week for our 6th graders: one during a music lesson, with the music teacher as the adult in the room, and one during math.

Protect the minutes

One thing I’d tell any school is to protect the minutes. We call it a 45-minute session, but in practice it isn’t. Some kids get their computer out and start right away, while others don’t begin until they’re prompted once or twice, and there’s always someone wrestling with a headphone or a login. Having an adult walk around, put a finger on it, and prompt keeps everyone moving.

That points to the biggest piece of advice I’d give. After COVID, a lot of schools invested heavily in online math programs, and too often teachers quietly hand the responsibility over to the platform, which is exactly why those programs fail. Schedule it so the adults in the room stay involved, don’t expect the kids to do it alone, and build in a rhythm of accountability. Even a simple milestone check every few weeks, where the adults in the building have to engage with how the kids are actually doing.

We’re definitely going to roll it out further

For me, the benefits of AI tutoring come down to scale and flexibility. Looking further ahead, I think an AI program could eventually build a really rich profile of a student, addressing both foundational gaps and grade-level content at once in a way most online math platforms simply don’t.

Now that the 7th and 8th grade content is available, we’ll be expanding Skye to our 6th, 7th and 8th graders next year, and using what we’ve learned about which students benefit most to be far more deliberate about how we group and schedule them.