AI tutoring for math intervention and MTSS

Where AI tutoring fits in each tier

• Tier 1 (core instruction for everyone): guided practice after the lesson, with feedback at each step, so students who are slightly behind get help before the gap grows.
• Tier 2 (targeted small-group support): a supplement to a teacher-led group, giving students extra practice on the exact skill the group is working on, on the days the group doesn't meet.
• Tier 3 (intensive, individualized support): at most a minor role. Students here need frequent time with a skilled adult, and any tool should be chosen by the team that knows the student.

Start with the skill, not the grade level

Intervention works when it targets the reason a student is stuck, which is often a skill from an earlier grade. A grade 9 student who can't solve equations may understand the idea of balancing perfectly well but trip over negative numbers or distribution, which come from grades 6 and 7. Sending that student more grade 9 equations practice won't fix it.

Use your screener and the student's own work to find the missing prerequisite, then point the tutoring there. A good tutor helps with this too: because it sees every step a student writes, it can show which step goes wrong, not just that the final answer is wrong.

Reading student work: a worked example

Here is a typical intervention problem and the most common wrong path. Asked to solve 3(x − 4) = 18, many students write 3x − 4 = 18 and get x = 22/3.

The issue isn't equation solving; it's distribution. The correct path is below, followed by the check that exposes the mistake.

Designing the dosage

How often and how long students practice matters as much as which tool they use. Tutoring that happens during the school day, on a set schedule, is more reliable than optional after-school time, because the students who most need it are often the ones who can't stay.

• Schedule sessions several times a week rather than one long block.
• Keep each session short enough to stay focused, typically 15 to 30 minutes for practice.
• Protect the time. If intervention is the first thing cut when a schedule gets busy, it won't work.
• Pair tutor practice with regular teacher contact, so a person checks in on progress at least weekly.
• Give it a fixed length, such as six to eight weeks, followed by a review of whether the student moves on, continues, or changes approach.

Progress monitoring with tutoring data

Tutoring data is useful for progress monitoring when it's tied to specific skills. Time spent alone doesn't tell you much. Look for skill-level information: which skills a student has shown they can do on their own, which mistakes keep coming back, and how much help they needed.

Treat this as one input alongside your screener and curriculum-based measures, not a replacement for them. Agree as a team on decision rules before you start, such as how many weeks without progress lead to a change in support. At the district level, reports should hide small groups so that individual students can't be identified.

Keeping people in the loop

• Teachers review the class view weekly and pull small groups around the most common mistakes.
• Interventionists see the same data the classroom teacher sees, so the student hears one consistent message.
• For students with IEPs or 504 plans, the student's team decides whether and how the tool fits, including accommodations such as typing instead of speaking.
• Families hear what their child is working on and how they can help at home.
• Students get a say too. Ask them what helps and what doesn't.

What Helio gives a teacher

• Assign a lesson to a few students.
• See who is stuck right now, and the common mistakes in a class this week.
• Check what each student understands, skill by skill, and replay any session.
• District reports show time on task and understanding for each school, with groups under 5 students hidden.