Brilliant teaches math by putting learners inside the problem. Lessons combine carefully sequenced questions, visual models, and hands-on problem solving so learners discover why a method works before relying on it. When they get stuck, Brilliant’s digital tutor, Koji, gives in-context guidance without taking over. Practice then removes support and brings ideas back in new settings, helping understanding become usable and durable.
Understanding starts with doing
Watching someone solve a problem can feel clear; solving it yourself is another matter. Brilliant asks learners to make decisions: move a point, test a pattern, estimate an answer, choose a strategy, or decide what must be true.
That effort is where learning happens. Research across STEM courses has found that active learning improves student performance compared with traditional lecturing.[1] Brilliant applies the same principle by making problem solving the center of the experience rather than an exercise added after the explanation.
Visual models make abstract ideas tangible
Math becomes difficult when symbols arrive before the idea makes sense. Brilliant starts with concrete diagrams, animations, computation, and objects learners can manipulate, then connects those experiences to formal notation.
A learner might drag a tangent line and watch its slope change, rearrange geometric pieces to reveal why a formula works, or adjust a graph and see how the equation responds. The interaction is designed to expose the mathematical structure that the symbols describe.
Lessons build one idea at a time
Brilliant lessons are carefully sequenced. Each section focuses on a manageable idea, begins with the simplest useful case, and increases the complexity as the learner’s understanding grows.
Brilliant often asks a question before presenting a procedure. That first attempt gives the explanation a purpose and surfaces what the learner already understands. But guided discovery is not the same as leaving someone to guess. Questions, examples, feedback, and direct instruction are arranged to keep the learner moving without doing the thinking for them.
The goal is more than completing problems. Learners should understand why a method works, recognize when it applies, and adapt it when the problem looks different.
Mistakes become useful information
A wrong answer can reveal a misconception, a skipped step, or a reasonable idea used in the wrong place. Brilliant’s interactive problems respond while the learner’s reasoning is still fresh, giving them a chance to revise rather than simply marking the attempt wrong.
This creates a safer kind of challenge. Learners can experiment, see what changed, and try again without treating confusion as evidence that they are “bad at math.” Timely feedback is especially useful when it answers three questions: Where am I going? How am I doing? What should I try next?[4]
Koji guides without taking over
In supported courses, Brilliant’s digital tutor, Koji, can see the problem, the interactive elements on screen, and what the learner has tried. Koji can ask a guiding question, highlight part of a diagram, break a problem into smaller steps, or offer another explanation.
The goal is not to answer questions faster. It is to help the learner find the next step and eventually continue without help. Koji offers more support while an idea is new and steps back when it is time for independent practice.
A digital tutor is still a tool, not a guarantee. If an explanation does not click, learners can ask for a different approach, return to prerequisite material, or get help from a teacher or human tutor.
Practice helps understanding stick
Understanding something once is not the same as retaining it or using it later. During independent practice, scaffolding, visual aids, and tutoring support fall away so the learner must retrieve and apply the idea independently.
Several design choices matter:
- Retrieval: Retrieving a method (rather than rereading it) strengthens later retention. [2]
- New contexts: A practice problem can change the surface details so the learner must recognize the underlying idea rather than copy a pattern.
- Spacing: Returning to important ideas across sessions improves long-term retention compared with concentrating all practice at once.[3]
- Interleaving: Mixing related problem types makes the learner decide which strategy applies. In mathematics, this can improve later performance even when blocked practice feels easier in the moment.[3]
- Feedback: Correcting a misconception while the reasoning is still visible helps the learner update the idea instead of rehearsing the error.[4]
Brilliant’s practice system uses learning history to select useful next problems. Review sets that combine material from preceding lessons are rolling out course by course and are reviewed by people before release.
What “understanding” looks like
A learner understands a mathematical idea when they can do more than repeat the steps. They can:
- Explain why a method works.
- Represent the idea with words, pictures, graphs, or equations.
- Choose an appropriate strategy without being told which one to use.
- Notice when an approach is failing and revise it.
- Apply the idea to an unfamiliar problem.
Brilliant is designed to build toward that kind of independence. It does not promise mastery from one lesson, and it is not a formal diagnostic assessment, school transcript, or complete replacement for every local curriculum. It can supplement school, support self-directed learning, provide enrichment, and help a learner work through a difficult concept.
Related resources
Sources
[1] Freeman et al., “Active Learning Increases Student Performance in Science, Engineering, and Mathematics”, 2014.[2] Roediger & Karpicke, “Test-Enhanced Learning”, 2006.[3] Rohrer & Taylor, “The Shuffling of Mathematics Problems Improves Learning”, 2007.[4] Hattie & Timperley, “The Power of Feedback”, 2007.
