Meet the Agentic AI Tutor: lessons that adapt to your answers

Most AI chat tools wait for you to ask something, answer, and then wait again. That can be useful, but it is not how a good tutor works. A good tutor has a plan for the lesson, checks what you understood, notices when you got something wrong and why, and explains it a different way before moving on. The Agentic AI Tutor in Studentlytica AI is built to work like that, using your own course materials.

This article explains what the Agentic AI Tutor does, what "agentic" means here, and why it matters for how you learn. It uses one example lesson throughout.

Two ways to use the Tutor

The Agentic AI Tutor has two modes.

Ask a question. You choose a course, type a question, and the Tutor answers from the materials you uploaded to that course: lecture notes, slides, handouts or your own notes. Every point carries a numbered source, and you can open the exact passage, with its page or slide, that it came from. If your materials do not cover the question, the Tutor says so instead of guessing. We explained this in detail in How the Studentlytica AI Tutor answers from your own notes.

Take a lesson. This is where the Tutor becomes an agent. Instead of answering one question, it takes you through a topic step by step: it plans, teaches, asks, listens to your answer and decides what to do next. The rest of this article is about lessons.

What "agentic" means

An AI agent is a system that works towards a goal over several steps, deciding at each step what to do next based on what has happened so far. Many AI products use the word loosely. For the Tutor, it means four specific things:

1. It plans. Before teaching anything, it reads your materials and proposes a short path of learning objectives for your goal.
2. It decides the next move. After each step it chooses what comes next: teach the next part, ask a question, explain again, ask you to explain in your own words, or sum up.
3. It adapts to your answers. A wrong answer leads to a different explanation, not the same one repeated. A right answer moves you on.
4. It suggests what to do afterwards. At the end it proposes next steps, such as flashcards on what you found hard, and does nothing until you choose.

Just as important is what the Tutor does not decide. It does not mark your answers by itself: multiple-choice questions are marked by code against the stored answer, so a correct answer is never marked wrong because the AI changed its mind. It cannot spend your allowance without your tap, and it cannot leave the lesson to do something else. An agent you can trust is one with clear limits.

An example lesson, step by step

The example below is an illustration of how a lesson flows. The course, the materials and the wording are made up for this article.

Akosua is a second-year biology student. Her course, BIOL 201, has her lecture notes and slides uploaded. She opens the Tutor, chooses Start a lesson, picks Learn a topic and types: how enzymes work.

1. The plan

The Tutor reads the relevant parts of her materials and proposes a lesson path, for example:

1. Describe what an enzyme does in a reaction
2. Explain the active site and how a substrate binds
3. Explain how inhibitors slow an enzyme
4. Describe how temperature and pH affect enzyme activity

Akosua already knows the first objective well, so she removes it, moves pH up the list, and saves the path. Nothing is taught until she starts. The plan is hers to change.

2. Teaching, then a question

The Tutor teaches the first objective in a few short paragraphs drawn from her lecture notes, with the page each point came from. Then it asks a check question:

A bar at the top of the lesson shows where she is: one segment per objective, and a line such as Objective 2 of 3: explain how inhibitors slow an enzyme.

3. A wrong answer, and a different explanation

Akosua chooses B. The answer is marked straight away: Not quite. The Tutor then does two things a good human tutor would do.

First, it gives its reading of what went wrong, clearly labelled as its reading and not a mark: It looks like you were thinking that an inhibitor destroys or uses up the enzyme. An inhibitor blocks the enzyme for a while; it does not use it up.

Second, it explains the idea differently. If the first explanation was a plain description, the second might be a worked example:

1. Picture one enzyme and a queue of substrate molecules.
2. An inhibitor sits in the active site, so the queue waits.
3. When the inhibitor leaves, the same enzyme carries on working.

The heading says what kind of explanation it is ("Explaining differently: worked example"), and the sources are still her own notes. Then she gets another question on the same idea.

4. Explaining it in her own words

This time she answers correctly, and the objective is marked as met. The Tutor then asks her to explain it in her own words. Putting an idea into your own words is one of the most reliable ways to find out whether you really understand it, and to make it stick. She types two sentences, and the Tutor responds to what she wrote.

5. The recap

When the path is done, or when Akosua ends the lesson, she sees what this lesson covered: which objectives she met, which need review, which she skipped and which were not reached. This comes from her answers, and it is not a grade.

6. Suggested next steps

Finally, the Tutor may suggest a few next steps, each on a card that says what it would use from her monthly allowance before she taps it. For example: make flashcards on inhibitors, make a short quiz, or plan a review session in the Planner. If she taps Not now, nothing happens.

Four kinds of lesson

When you start a lesson, you choose what kind of help you want:

• Learn a topic. A short lesson on one topic, taught step by step, with a question after each part. This is the example above.
• Help with a problem. Paste a problem from your course. The Tutor gives hints and asks guiding questions, one at a time; you do the steps. The next hint is one tap away when you need it, and the full worked solution stays out of the way until you ask for it. Please do not use this for an exam or assessed work you are doing now.
• Fix my weak spots. The Tutor looks at your recent answers to practice questions in a course, picks the one to three topics where you did worst, and plans a lesson on them.
• Exam practice. Questions first, mixed across up to six of your course's topics, with weak topics and topics due for review first. After a wrong answer you get one short explanation, then another go.

Why this matters for your learning

Decades of research on tutoring point to the same few ingredients. A tutor helps most when it works step by step, checks understanding often, gives feedback on each step, and adjusts to the learner. You can read more in What is an intelligent tutoring system?.

Recent studies also show the risk of the opposite. When students use a general chat assistant that simply gives them answers, their practice work can look better while their later exam results get worse, because the assistant did the thinking for them. We summarised that research in AI tutors: what works, and what can go wrong.

The Agentic AI Tutor is designed around that evidence:

• It makes you do the thinking. Questions after each part, hints before solutions, and explaining in your own words.
• It responds to your mistakes. A wrong answer gets a diagnosis and a different explanation, not a repeat.
• It stays with your course. Explanations come from your own materials, with sources you can check, so you learn what your lecturer actually taught.
• It remembers. Your answers update your progress on each topic in the course, which the Tutor uses for weak-spot lessons, exam practice and your daily study brief.
• It is honest. Marks come from code, the Tutor's reading of a mistake is labelled as a reading, and the recap is not a grade.

No tool can promise better grades. What a good tutor can do is make your study time more active, and help you find and fix what you do not yet understand.

Practical details

• Allowance. Starting a lesson uses one lesson from your monthly allowance. Replies inside a lesson, hints and the recap use nothing more. Each lesson has a step limit, and the page tells you when only a few steps are left. See Pricing for how many lessons each plan includes.
• Your materials. Lessons need at least one ready material in the course. Exam practice also needs the course's topic map.
• Privacy. Your materials and lessons are private to your account. To write each step, Studentlytica AI sends our AI provider only the passages needed for that step and your answer, never your name, email address or account details. Your materials are not used to train AI models. The Privacy Notice has the details.

Try it

Open the Agentic AI Tutor, choose Start a lesson, and pick a topic you are studying this week. Change the plan if it is not quite right, answer honestly, and read the sources when an explanation surprises you. Treat the Tutor as a guide to your materials, not a replacement for them.