Students independently work through Khan Academy's instructional videos and practice problems, submitting answers that the platform marks right or wrong. Where a school system or school has purchased it, students can also use an optional chat panel to ask the AI tutor, Khanmigo, for help on a problem before or after they submit, ask where to start, or bring questions and content from outside Khan's courses. Students can finish their work without ever interacting with Khanmigo. In school systems using Khan's interim assessments, some items in the assessment require students to type out an explanation of their thinking in response to Khanmigo's questions.

For student tutoring, AI provides support and feedback while students are working on questions. The Khan Academy platform underneath it provides human-written problems, scores them, tracks mastery, and serves videos without AI. While students are working on these problems, they can engage in conversation with Khanmigo, which uses generative AI to respond, rather than pulling from pre-written hints. Behind the scenes, during the conversation, Khan passes the problem, the step-by-step solution, and what the student has already mastered and recently gotten wrong to the AI model, which helps make the AI response more accurate and tailored to the student. The chat also displays suggested questions that students can click if they do not know what to ask. Khanmigo acts differently before a student submits an answer versus after they submit an answer. Before submission, Khanmigo gives hints and strategies but will not provide the answer. After the student submits, Khanmigo gives more step-by-step guidance, and sometimes the answer, so students can find where they went wrong. During interim assessments, for some items, students complete a more traditional multiple choice or fill-in-the-blank assessment item and then engage in a short conversation with the AI in which the AI asks them to explain their answer. Behind the scenes, the students' responses are scored by the AI based on a rubric written by Khan’s content authors.

For teachers, Khanmigo works through the teacher dashboard: it summarizes student chat logs, reports how a class is progressing, suggests what to focus on, and proposes student groupings with a recommended assignment for each. There are also numerous teacher tools to help draft lesson plans and complete other administrative tasks. Teachers assign practice aligned to the lesson they are teaching from their own core curriculum, and monitor student progress during class. Teachers can also assign personalized practice to fill gaps from prior years or move students ahead. In the rotation model Khan designs for, teachers lead small-group instruction with one group while another works on Khan. They can read students' chat logs with Khanmigo, accept or override the groupings and assignments it proposes, and use its free planning and administrative tools. Teachers also receive flags from the system if it detects inappropriate conversations with Khanmigo. Teachers can then view the chats and decide whether to discuss the interaction with the student.

Leaders can purchase student access to Khanmigo, either through a school system agreement or by school, which determines whether students can use the tutor at all. They set the instructional design Khan sits inside: which core curriculum teachers use, when practice happens in the schedule, and whether Khan serves on-grade-level practice, intervention, enrichment, or all three. Leaders also have access to dashboards that allow them to track usage and learning on the site and receive warnings about inappropriate or concerning conversations with Khanmigo.