AI writing analysis
Our KLX model analyzes a text for signs of AI-generated writing and reports the uncertainty of the result.
Data and AI technologies
Modular software for academic work
ModulesIn use · Turkish In use · Turkish In use · Turkish Planned Planned Planned
Generative AI has changed how academic work is written and assessed.
Each module focuses on a specific task and shows the basis for its result.
Annotated report
Explains why a passage was flagged and shows the confidence level and the limits of the analysis.
Reference list generator
Creates an APA 7 reference list from the source details you enter.
Reference and citation check
Checks whether cited sources exist and whether in-text citations match the reference list.
AI-use declaration
Creates an AI-use declaration from the information you provide. Runs in the browser without sending the entered text to a server.
Text similarity check
Compares a text with published sources and shows where passages overlap.
ProductsIn use · Turkish
In use in Türkiye. Expanding to English and Arabic.
TezMakale
Our modules are in use on TezMakale.
1,400+anonymous users
300+registered users
20+academic researchers
October 2026 · TezMakale records
Visit TezMakale.comKLX
KLX is our own machine learning model for Turkish.
Machine learning
It learns from Turkish academic texts written by people and by AI. We keep measuring and improving it with new data.
Built on Turkish
It is grounded in the grammar of Turkish and the conventions of academic writing. It is not a model built for another language and adapted to Turkish.
Built in Türkiye
We develop it in Türkiye with our own team. No outside AI model or service is used in Turkish analysis.
Sample report
The sample text is fictional. The explanations were written by hand and are not model output. Select a flagged sentence to see its explanation.
Illustrative example · not an actual analysis
The timing of feedback plays a pivotal role in developing students' writing skills. In this study we compared how quickly feedback was returned after draft submissions in two undergraduate courses. Students who received feedback within a week made fewer formatting errors in their second drafts than those who received it later. This finding clearly demonstrates that feedback is an indispensable part of the learning process and has a transformative impact on education. Because the sample was small, the result should not be generalized to other courses.
Two sentences are flagged in this example. The flags identify passages to review. They do not establish that the text was generated by AI.
Why it was flagged
Confidence
Limitations
Method
We are open about how we measure.
Evaluation method
We test the accuracy of AI writing analysis tools, starting with our own KLX model. We document the protocol before we examine the test results.
Evaluation by language
We evaluate performance separately for each language. Arabic and English are next.
Interpreting results
An analysis result alone does not prove that a text was generated by AI. Results should be considered alongside human judgment.
Company
Co-founders
