Pillars Idea Process Blog Contact DE
Learning Innovation with Flipped Tutoring.
GenAI tutor for the self-study phase.
Just watching videos. A self-study phase that finally engages.
Student co-design Adaptive paths Deeper engagement
L
Learning
Innovation
FFFFFFFFFFFFFFFFFFFFFFFFFFFF
Flipped
T
Tutoring

P–01

Co-designed with students at every step, not just handed to them to consume.

Students aren't passive recipients. Co-design workshops, usability tests, and continuous feedback loops put their lived experience at the center of every design decision. They shape the tutor before it launches and keep refining it through use.

P–02

Paths that adapt. Intelligence that fits.

No two students start from the same place. The AI reads where each learner actually is, adjusts knowledge checks to what they get wrong, and reshapes the path as understanding grows. Questions get answered in natural language, mid-study. Human tutors review every generated check before it goes live. And the students using it keep shaping it, their feedback closes the loop between what the GenAI tutor offers and what learners actually need.

P–03

Learning that goes deeper.

The self-study phase is where most students get lost, and most learning gets lost with them. Pre-recorded videos and course materials sit there, static. Students can watch and read, but they can't ask. Flipped tutoring with GenAI changes that: the material becomes something you can have a conversation with. Students arrive prepared, go further in class, and learn to have a dialogue with what they study, not just sit with it.

P–01 · Pillar 01 / 03
Co-Design

Co-designed with students at every step, not just handed to them to consume.

Students aren't passive recipients. Co-design workshops, usability tests, and continuous feedback loops put their lived experience at the center of every design decision. They shape the tutor before it launches and keep refining it through use.

P–02 · Pillar 02 / 03
Adaptive AI

Paths that adapt. Intelligence that fits.

No two students start from the same place. The AI reads where each learner actually is, adjusts knowledge checks to what they get wrong, and reshapes the path as understanding grows. Questions get answered in natural language, mid-study. Human tutors review every generated check before it goes live. And the students using it keep shaping it, their feedback closes the loop between what the GenAI tutor offers and what learners actually need.

P–03 · Pillar 03 / 03
Engagement

Learning that goes deeper.

The self-study phase is where most students get lost, and most learning gets lost with them. Pre-recorded videos and course materials sit there, static. Students can watch and read, but they can't ask. Flipped tutoring with GenAI changes that: the material becomes something you can have a conversation with. Students arrive prepared, go further in class, and learn to have a dialogue with what they study, not just sit with it.

Scroll
The case for GenAI in Flipped Tutoring

Videos are
not dialogues.

Imagine if Plato only had videos of Socrates.

The flipped classroom moves lectures out of the room so deeper learning can happen in it. What goes wrong is what happens in between: students watch recordings and read materials alone, no feedback, no one to ask. We replace that gap with a GenAI tutor that can hold a real dialogue, listening, responding, and adapting, keeping students moving, motivated, and engaged, so they arrive ready for what the live session is actually for: discussing with peers and mentors, questioning ideas, and going deeper together.

AI-generated image: Plato watching a video of Socrates, who has asked a question, but Plato cannot reply back
Process · 05 Phases

Five steps to bridge this gap.

01
Investigate
We examine the cognitive and metacognitive challenges students face when learning with GenAI. Through a systematic literature review, expert interviews, and a pilot study with students, we identify what skills are needed to critically evaluate AI outputs and integrate them into learning. These findings directly shape the tutor's design.
AP 1
02
Develop
We redesign the self-study phase from scratch. Existing learning materials are analyzed for gaps, a framework for adaptive learning paths is built, and static assignments are replaced with interactive knowledge checks. Students test every iteration and their feedback reshapes the paths before the next cycle begins. Everything is documented as OER as we go.
AP 2
03
Build
We develop the AI tutor across three prototype cycles, each adding capability, each evaluated before the next begins. Prototype 1 establishes the platform and basic GenAI LLM model integration. Prototype 2 adds generative knowledge tests. Prototype 3 introduces fully adaptive learning paths. Students co-design throughout, from usability tests to feedback loops.
AP 3
04
Evaluate
Two structured evaluation rounds. The first measures how students experience adaptive learning paths compared to traditional formats. The second tests the refined system for learning success, usability, and acceptance. Both combine qualitative methods, interviews and focus groups, with quantitative data: usage analytics and learning progress analysis. Results feed directly back into the tutor.
AP 4
05
Share
Everything we build is released as Open Educational Resources. The platform architecture, generative knowledge tests, adaptive learning paths, and the full AI tutor system are documented and published so other modules, faculties, and institutions can adopt and extend them.
AP 5
Let's
the classroom

Researchers, students and faculty welcome. Get in touch to follow the project, join a co-design workshop, or collaborate on the OER release.

[email protected]