Clastify logo
Clastify logo
Exemplars
Review
HOT
Tutoring
Click here to explore thousands of examiner marked IA, EE and TOK exemplars!

How Clastify Uses AI to Improve Marking

Janek

By Janek

10 Jun 2025

How Clastify Uses AI to Improve Marking cover image
Get feedback on your IA
Marked only by official IB examiners

At Clastify, we take great pride in the quality and accuracy of our coursework reviews and mark schemes. All marking/comments in our reviews and mark schemes are written by experienced IB examiners, but to make the feedback even more effective, we occasionally use AI to assist with clarity - never to replace expert judgment. In this post, we'll explain what parts of the review are written by humans, where and how AI is used, and show examples of comments before and after AI augmentation.

 

 

How Clastify Uses AI to Improve Marking

 

 

 

What’s Done by Real People (and How AI Helps)

 

When your IA, EE, or TOK submission is reviewed (or a mark scheme created) by Clastify, here’s what actually happens:

 

  1. A certified IB examiner carefully reads the work and evaluates it according to IB criteria.
  2. For each subcriterion, the examiner:
    1. Marks it as correct, incorrect, or not applicable
    2. Highlights the relevant parts in your work
    3. Provides a short explanation or note
  3. Based on these human inputs, our AI then suggests a clear, student-friendly comment to explain the issue or offer guidance. The goal is to transform the examiner’s expert feedback into a comment that’s easy to understand and actionable
  4. However, every AI-generated comment must be manually approved or edited by the examiner. Nothing is added to your review/mark scheme without the reviewer’s approval.

 

It’s important to emphasize: all grading and substantive feedback is entirely done by human, IB-certified examiners. AI is only used to improve the clarity and tone of the comments — never to make judgments.

 

 

Real Examples

 

To help you understand how Clastify’s review system works in practice, here are a few real examples of how our human reviewers and AI collaborate - always with the human in full control.

 

Example 1:

 

Subcriterion: The student clearly states the independent and dependent variables in the research question or two correlated variables (if applicable)

 

What the examiner did: marked as incorrect, highlighted the research question, and wrote a quick note:

 

units missing

 

Based on that the AI generated the following comment:

 

The student does not specify the dependent variable clearly in the research question. To improve this, it is recommended that the student state how the activity of salivary amylase will be measured, such as by indicating the rate of starch hydrolysis in mg/min or another appropriate unit. For example, the research question could be revised to "How does varying the pH levels (0, 2, 5, 7, 10, and 14) affect the rate of salivary amylase activity (mg/min)?"

 

The reviewer read the AI-suggested comment, determined that it's accurate, and chose to accept it.

 

Example 2:

 

Subcriterion: The document includes a conclusion section based on the results of the data analysis

 

What the examiner did: marked as correct, highlighted the first paragraph of the conclusion.

 

Based on that, the AI generated the following comment:

 

The document includes a conclusion section that is based on the results of the data analysis. It summarizes the findings regarding salivary amylase activity at different pH levels and their implications, aligning with the results of the data analysis. However, to fully meet the criterion the student should include more data from the "Raw Data" section.

 

The reviewer read the AI-suggested comment and decided that it doesn't fully reflect his judgment. Therefore, the reviewer decided to make manual edits, which resulted in the final version:
 

The document includes a conclusion section that is based on the results of the data analysis. It summarizes the findings regarding salivary amylase activity at different pH levels and their implications, aligning with the results of the data analysis. Therefore, it meets the criterion for including a conclusion section based on the data analysis.

 

 

How AI resulted in an increase in quality

 

At Clastify, we mark thousands of IB IAs, EEs, and TOKs every year. One of the biggest challenges we faced before implementing AI was consistency and clarity in our comments.

 

Our IB examiners are highly qualified, but what seems clear to one person might feel vague or confusing to another - especially when students are interpreting feedback under pressure. While the content of the reviews was always strong, we knew that the way it was communicated could be improved.

 

With AI now assisting in comment augmentation, we’ve seen significant improvements in:

 

  • Clarity: AI helps turn quick notes from reviewers into clear, student-friendly feedback.
  • Consistency: Terminology and phrasing are now more standardized, so students get equally high-quality comments across subjects and reviewers.
  • Tone: AI ensures comments come across as constructive and encouraging, even when the feedback is critical.
  • More examples: AI ensures that the examiner comments always contains specific examples showing how the feedback could be implemented - something that our users find very helpful.

 

Since introducing AI augmented comments, we’ve seen:

 

  • A 34% increase in users who rated their review as 5 stars
  • A 21% drop in follow-up clarification emails asking What does this comment mean?
  • A 24% decrease in users reporting mark scheme subcriteria as Confusing

 

All of this has been achieved without compromising on accuracy - every comment is still reviewed, edited, or approved by a certified IB examiner before being published.

 

 

Use of AI and confidentiality

 

At Clastify, protecting coursework submitted for review is one of our top priorities. Your files are never made public or shared outside of Clastify. Only the IB examiner assigned to your review (and our internal team, where strictly necessary) can access your full document.

 

When AI is used, it only works with the examiner’s notes and highlights to make feedback clearer. While we do use external AI models, these models do not retain any data. Your coursework always stays private and secure within Clastify.

 

 

Have questions?


If you would like to learn more about how we integrate AI to augment comments or have any questions, feel free to email us at review@clastify.com. We're here to help!

 

At Clastify, we’re committed to combining the expertise of real IB examiners with the clarity and consistency that AI can provide — so that you get feedback that’s not only accurate but genuinely helpful. We're always looking for ways to improve, and we believe this human–AI collaboration is a major step forward in making our reviews clearer, fairer, and more actionable for every student.