Feature Request: Personalized Curriculum-Based Exercises for Students

[pr-elhajji]

Is your feature request related to a problem? Please describe.
As a student, I want OpenTutor AI to go beyond being just a chat interface like Open Web UI. Currently, it doesn’t tailor its responses to my specific educational needs or school curriculum, which makes it feel generic and unhelpful for my studies. I’m frustrated that it doesn’t act like a teacher who understands my academic program, age, and personal learning requirements.

Describe the solution you'd like
I’d like OpenTutor AI to propose exercises and content that align with my school curriculum and my specific needs as a student. It should take into account factors like my age, grade level, and the subjects I’m studying, functioning more like a teacher. For example, it could ask me about my curriculum (e.g., “What subject are you studying?” or “What’s your grade level?”) and then suggest relevant exercises or explanations. The system should adapt its tone and complexity based on my age and provide structured guidance, not just open-ended chat responses.

Describe alternatives you've considered
One alternative is to manually input my curriculum details each time I use the AI, but this would be tedious and impractical. Another option is to have predefined curriculum templates (e.g., for common grade levels or regions) that I could select from, though this might not fully address my unique needs. A third idea is to integrate with existing educational platforms or syllabi, but that could be complex and limit flexibility.

Additional context
This feature would transform OpenTutor AI into a true educational tool, making it more valuable for students like me. It could start by supporting basic customization (e.g., age and subject selection) and later expand to integrate with specific curricula or learning goals. The focus should be on mimicking a teacher’s ability to guide and challenge me with relevant exercises, rather than just responding to random questions. This would require the AI to store some user-specific data (like age or curriculum) and use it to personalize the experience.

[pr-elhajji]

Let's look at csfirst.withgoogle.com for design ideas.

[TarekAi]

in my opinion, the main features that a student should have in the left menu of Open TutorAI should provide easy navigation and access to essential learning tools (// maintain a user-friendly and structured interface). The first feature is a Dashboard which will provide an overview of the student’s progress, upcoming assignments, recent activities and some recommendations (AI or teacher generated). Then, My Classrooms , a space where students can access the classrooms created by their teachers, interact with AI tools assigned to them, and follow structured and personalized lessons.
Assignments where students can view, complete, and submit assignments and quizzes assigned by their teachers ...
Progress & Reports a performance tracking feature where students can monitor their learning progress, view detailed reports, and receive insights on areas for improvement ..
Messages a communication space for interacting with teachers or collaborating with classmates.
Profile a space where students can personalize their learning preferences

What do you think ..?

[TarekAi]

https://discord.com/channels/1344744138020622376/1353344593499914312

[XredaX]

I love the idea of adaptive, curriculum-based exercises. In my opinion, adding a dynamic difficulty ramp that adjusts based on real-time student performance could help prevent frustration while still offering a challenge

[musaabhasan]

I would model this as curriculum-grounded practice generation, not only as a richer chat profile. A practical first version could have four layers:

1. Curriculum map: subject -> grade level -> unit -> learning objective -> prerequisite objectives. Even a lightweight JSON/YAML map would be enough to start, and schools could later replace it with their own syllabus.

2. Learner profile: age/grade, preferred language, accessibility needs, target curriculum, and current mastery per objective. I would avoid storing broad personal data unless it directly changes instruction or assessment.

3. Exercise generator with constraints: instead of asking the model to make an exercise, pass the exact objective, difficulty, expected misconception, output format, answer key, and rubric. For example: objective = fractions addition, level = developing, misconception = denominator addition, output = 3 scaffolded items + worked solution + hint sequence.

4. Teacher validation loop: generated exercises should have a draft status until a teacher approves, edits, or rejects them. Student performance can update mastery, but the curriculum/objective alignment should remain auditable.

For adaptive difficulty, I would use a simple mastery state first: not started, emerging, developing, proficient, extension. Move learners up only after repeated success across varied items, and move them down only when errors repeat on the same prerequisite. That is more defensible than changing difficulty after one right/wrong answer.

A strong MVP could be: select curriculum + grade + subject, choose one learning objective, generate a short diagnostic, then produce a personalized practice set with hints and a teacher-visible rationale explaining why each item was selected. That would make the feature useful for students while still giving educators control over quality and alignment.