CK-12 Flexi AI Tutor
An answer begins the interaction. Clarification, source choices, and a next challenge give students ways to keep learning.
- An AI tutor can answer the question and still skip the learning. Students need help getting unstuck without the tutor doing the work for them.
- I led UX for Flexi with product, engineering, and curriculum partners: follow-up actions, so a student can ask for the next kind of help without writing a better prompt, and labels that separate AI-generated answers from CK-12 Library content.
- In an external six-week study, teachers reported more student engagement and confidence, with prompting and reading level as the barriers. In CK-12’s analysis of Flexi in Adaptive Practice, answer-seeking questions fell from 72% to 52% by a student’s eighth question.
On this page
Flexi is CK-12's AI student tutor. As Lead UX Designer, I worked with Product, Engineering, Data Science, curriculum, and research partners. I turned the tutor's capabilities into interactions students could use.
My goal: help students past confusion and keep the learning effort with them. CK-12's platform reach does not measure this feature's impact.
Refusing help can drive a student away. A bare answer turns the tutor into a shortcut. The middle path: help now, then a clear way to clarify, practice, or go deeper.
A blank prompt assumes a student knows how to ask for better help. The interface shows follow-up actions instead: simplify, add detail, use an analogy, translate, rephrase, or try a challenge.
The tradeoff: a menu asks a confused student for one more decision. So each choice must be clear.
The interface labels AI-generated content apart from CK-12 Library explanations. Students can ask for a different one. A source label is not a confidence estimate, but it helps the student orient.
Teachers need to know whether help supports learning or replaces it. This case study covers the student side. The Foresights and Insights case study covers teacher analytics.
I did not author this research. A Spring 2025 Flexi study(opens in a new tab) by Leanlab Education, an external nonprofit, followed ten middle- and high-school teachers in five states over six weeks. Teachers reported more student engagement, confidence, and curiosity. They also named barriers: students struggled to write prompts, and some teachers worried about over-reliance. Leanlab calls these early indicators, not conclusive.
In a CK-12 comparison(opens in a new tab), nine of ten teachers preferred Flexi to Khanmigo and Quizlet's Q-Chat. All ten named its visual, interactive responses as an advantage.
The follow-up actions address that prompting barrier.
An explanation, its source, and the next learning action belong together. Split them, and recovery from a weak answer takes more work.
The image at the top shows the follow-up menu beside the conversation.
CK-12's analysis of student dialogue(opens in a new tab) sampled 5,000 students and studied 15 cases closely. Students who grew asked follow-ups. Students who did not tended to copy and paste.
In Adaptive Practice, Flexi offers "Give Me a Hint" and "Clarify This" and does not reveal the answer. There, direct answer-seeking questions fell from 72% of students' first questions to 52% by the eighth. Learning-oriented questions rose from 11% to 26%.
The analysis is descriptive. It does not show that my design caused a learning gain. It does support designing for conversation quality over length.
I would test follow-up choices by asking students to explain or apply what they learned. The question is whether they can proceed on their own.