Students individually work through a set of math problems, aligned to the standards or skills they’re targeting, that their teacher has selected. They're required to show their thinking before they can submit an answer—by typing, handwriting, or drawing on the screen; dragging and dropping; marking up the problem; recording a spoken explanation; uploading a photo of work done on paper; or using on-screen manipulatives like number lines and protractors. When students get a problem incorrect or get stuck, they can ask for a hint about where they went wrong. They can switch the problem and hints into their home language or have them read aloud. After working on their own, students can look at classmates' strategies and talk through the differing approaches to the same problem.

The AI analyses to grade each student's response and pinpoint any misconceptions or errors; then, it generates a hint, escalating to a more detailed hint if the same mistake recurs. It accepts mathematically equivalent answers (e.g., a decimal, fraction, or percentage that means the same thing) and overlooks minor formatting like spacing, unless the question calls for a specific format. It saves every attempt for the teacher, including those a student erases or redoes. Across the class, it flags where students are struggling, surfaces common incorrect answers, and turns that student work into discussion prompts the teacher can use. It can translate problems and hints into 140+ languages and read them aloud.

Teachers build assignments by selecting problems from Magma's standards-tagged bank, choosing the standards or skills to target and the level of challenge. They can assign different sets to different students to match where each is. Teachers decide where the assignment fits into their math block: as a warm-up, a formative check, independent practice, or a quiz. While students work, teachers can view a live heat map that shows, problem by problem, who got it right on the first try, who got it eventually, and who hasn't gotten it yet. From there, teachers can freeze the class mid-assignment and use the live activity dashboard to pull up and anonymously display students' solutions on the whiteboard—selecting and sequencing several to spark a whole-class discussion comparing approaches, drawing on the AI's flagged common errors and discussion prompts. Teachers can also use the live heat map at the end of an assignment to decide what to reteach. They can also replay any student's attempt step by step to see how the student's thinking unfolded and review the AI's grading and override it when it's off.

School and school system leaders adopt Magma across classrooms or a whole system and decide how it's used—from regular formative checks like do-nows and exit tickets to a school system-wide common assessment in place of some interim tests. Using a school system dashboard, leaders can build and push out assessments to classrooms, see results in real time, and compare performance across schools, classes, grade levels, and standards to find where students are struggling system-wide. They can also provide implementation support, such as running professional learning communities around Magma data.