Higher Education

AI your students can use without it writing their work.

A tutor that questions rather than answers, taught from your own module materials, with analytics that show where understanding breaks down.

The challenge

Students already use AI. The question is which kind.

General chatbots give answers and invite misconduct. Banning them does not stop their use. Universities need an AI that builds the reasoning assessment is meant to test, and evidence of what students understood.

How universities use it

Four learning modes for academic work

  • Socratic Partner

    Seminar and essay preparation

    The tutor asks questions only. It probes assumptions and asks students to justify their claims, so they arrive at the seminar having done the thinking.

    I think the policy failed because it was underfunded.

    What evidence would show that funding, rather than design, was the cause?

    Illustration with synthetic data.
  • Academic Feedback

    Feedback on a student’s own draft

    Students paste their own writing and get criterion-referenced feedback against the marking criteria. The tutor gives feedback; it does not rewrite.

    Can you check my introduction against the criteria?

    Your aim is clear. The criteria ask for a stated scope; which part of the topic will you leave out?

    Illustration with synthetic data.
  • Gibbs Reflection

    Reflective writing and placements

    The tutor takes students through Gibbs’ cycle one phase at a time, from description to an action plan, and keeps the reflection focused.

    The patient handover went badly.

    Description first: what happened, in the order it happened?

    Illustration with synthetic data.
  • Adaptive Learning

    Independent study between sessions

    The tutor reads how a student is doing on every turn and changes its approach to suit. It draws diagrams and sets worksheets from the module materials, then marks the answers.

    There’s so much in this week’s reading. I don’t know where to start.

    Let’s take one step. Which of the three models in the reading did your lecturer spend longest on?

    Illustration with synthetic data.

All eight learning modes

Academic integrity

It is built not to write assessed work.

When a student asks it to write, rewrite or polish assessed work, the tutor declines with a fixed reply and turns the request back into learning. Originality is scored on every turn, so staff can see whose thinking is on the page.

Read about integrity in the trust centre

Write my introduction for me.

I can help you plan, structure, and improve your own work, but I can’t write assessed sections for you, or help present that as your own work. Share your own draft, or tell me what you’re trying to say, and I’ll help you build it yourself.

The tutor’s actual integrity reply. Learner message is an illustration.

From students

What postgraduate students say

Rather than giving me the answers, it challenged my thinking like an academic supervisor, prompting me to think more deeply and articulate the reasoning behind my decisions.

PhD studentUniversity of Greater Manchester

My experience with Nested Learning strengthened my employability and helped me secure my first full-time role. I would strongly recommend it.

MSc studentLeeds Beckett University alumni

Fits your setting

Works with what you already have

  • Your module materials

    Upload handbooks, marking criteria and readings as rulebooks. The tutor teaches from those.

  • Your virtual learning environment

    A SCORM package lets the tutor sit inside your existing VLE course pages.

  • Evidence for staff

    Module leads see confidence, originality and concepts for their learners, and a summary of every session.

  • Voice and languages

    Students can talk to the tutor instead of typing and have replies read aloud, in British English. The tutor replies in 13 languages.

Research

What research on AI in higher education found

A randomised trial of AI tutoring, and a study of how students' agency with AI develops. Peer-reviewed studies of AI in education in general, not of Nested Learning.

Randomised controlled trial University students

AI tutoring outperforms in-class active learning: an RCT introducing a novel research-based design in an authentic educational setting

Kestin, Miller, Klales, Milbourne and Ponti (2025). Scientific Reports 15, 17458

Students learned more, in less time, from an AI tutor built on research-based teaching practice than from an in-class active-learning session, and reported higher engagement and motivation.

What it does not show: It studied a different AI tutor, not Nested Learning.

Read the paper: AI tutoring outperforms in-class active learning: an RCT introducing a novel research-based design in an authentic educational setting

Longitudinal multiple-case study over seven weeks 6 university English learners (3 focal cases)

Understanding development of learner agency in GenAI-assisted oral argumentation through the critical GenAI literacy-based dialogic teacher feedback

Li, Chen and Sun (2026). Studies in Higher Education (published online 12 October 2026)

Learners got better at prompting generative AI, but their ability to judge the content and logic of its output developed less consistently. Learner agency developed “in differentiated and nonlinear ways”, and sustained, responsive teacher feedback mattered.

What it does not show: Six learners in one English-language setting; qualitative, so it shows patterns rather than effects. Not a study of Nested Learning.

Read the paper: Understanding development of learner agency in GenAI-assisted oral argumentation through the critical GenAI literacy-based dialogic teacher feedback

Questions

Questions from universities

Can universities use an AI tutor without encouraging academic misconduct?

Yes, if the tutor is built to teach rather than to write. The Nested Learning AI tutor declines to write, rewrite or polish assessed work, redirects the student to their own reasoning, and scores originality on every turn so module staff can see whose thinking is on the page.

Does the AI tutor work inside our virtual learning environment?

Yes. A SCORM package lets the tutor sit inside existing VLE course pages, and it teaches from the module materials you upload, such as handbooks, marking criteria and readings.

Which learning modes suit Higher Education?

Four suit Higher Education especially well: Socratic Partner for seminar and essay preparation, Academic Feedback for feedback on a student's own draft against the marking criteria, Gibbs Reflection for reflective writing and placements, and Adaptive Learning for independent study.

What can module leads see?

Module leads see confidence, sentiment, originality and the concepts covered for every turn, plus a short summary of every session: main topics, strengths, difficulties and next steps.

Is there evidence that AI tutoring helps university students?

Published research is encouraging. A 2025 randomised controlled trial in Scientific Reports (Kestin and colleagues) found university students learned more, in less time, from a research-based AI tutor than from an in-class active-learning session. That study tested a different AI tutor, not Nested Learning.

How do we start a pilot?

Book a call in our booking calendar. A pilot runs on your own module materials, with your students.

Sign in

Choose where you want to go.

Learning analytics opens on its own site.

Learn About Nested Learning

Preview: coming soon

Soon you will be able to learn about Nested Learning by using it: our own AI tutor will answer your questions about the product, the learning modes and pilots.

How would this work for my apprentices?

Good question to start with. Which part of their standard do they find hardest to practise outside work?

Illustration of how it will look. Not a live conversation.

In the meantime, book a pilot call and we will show you the tutor live.