From the Field to Technology: Amanin's Experience Implementing AI Proctoring
Building an Artificial Intelligence solution for examination monitoring is not simply a matter of developing a system that can recognize faces or detect suspicious activities.
The real challenges emerge when the technology is deployed in real-world situations. Participants use different devices. Internet connection quality is not always consistent. The number of participants can reach thousands. Proctors have to manage many participants simultaneously. Participants may encounter technical difficulties. And when potential violations occur, examination organizers need evidence that can be reviewed afterward. This is where implementation experience becomes extremely important.
For Amanin, Proktor Amanin did not emerge all at once in the form it has today. The product has evolved through a long process: starting from real user needs, being tested in various scenarios, receiving feedback from customers, and continuously being refined based on field experience.
To date, Proktor Amanin has been used by various institutions, including Institut Teknologi Bandung (ITB), IPB University, Universitas Indonesia (UI), Telkom University, ACI, Indomaret, Schoters by Ruangguru, and a number of other organizations.
Each implementation has provided different lessons. And it is precisely through these experiences that Amanin has learned how AI proctoring should be built.
Starting with One Fundamental Question: How Can Online Examinations Be Made Trustworthy?
When examinations move into an online environment, a fundamental question arises: How can we ensure that the person taking the examination is truly the registered participant and is completing the examination honestly?
At the beginning of Proktor Amanin's development, this need came from universities seeking to improve the security of online examinations. Benchmarking existing solutions showed that there was still room to enhance the use of AI, particularly in identity verification and participant monitoring. Based on this need, Amanin began developing facial recognition and various AI-based monitoring capabilities. However, throughout the development journey, Amanin discovered that identity verification was only the beginning of the problem.
A participant could pass facial verification but subsequently be replaced by another person. There could be a second person nearby. A participant could open another application, take a screenshot, or access a prohibited website. Therefore, Proktor Amanin evolved into a system that does not merely verify who the participant is, but also monitors what happens during the examination.
From Facial Recognition to Activity Monitoring
Proktor Amanin has developed a number of capabilities to monitor participants, including facial recognition, multiple-face detection, browser and window activity monitoring, copy-paste detection, screenshot or screen-recording detection, as well as both AI-based and human-based proctoring. This approach brings an important change to the monitoring process.
While manual monitoring depends heavily on a human's ability to observe each participant, AI can help monitor large numbers of participants consistently and simultaneously. The system can also record activities and provide data that can be reviewed afterward.
As a result, proctors do not have to rely solely on memory or momentary observations. They have a digital trail of the examination process.
Learning from Different Types of Implementations
One of the strengths of Amanin's experience is that Proktor Amanin has not been used in only one type of environment. University entrance examinations have different requirements from student examinations. Corporate assessments differ from placement tests at educational institutions.
Therefore, every implementation provides a different perspective on how AI proctoring should work.
1. ITB: Learning to Handle Large-Scale Examinations
ITB has been one of the important users in the journey of Proktor Amanin. Proktor Amanin has been used to manage new student admissions examinations through the ITB Independent Selection and the International Undergraduate Program. This experience demonstrated that the system could operate at large scale. According to ITB's testimonial, the system has been used to manage more than 7,500 participants in a single examination session without significant problems.
The system also operates with relatively low bandwidth requirements and does not require high-specification devices. In the 2025 Independent Selection, the implementation experience once again demonstrated the importance of automated identity verification and large-scale monitoring capabilities.
In its testimonial, ITB stated that a single proctor could supervise up to approximately 50 participants simultaneously, supported by AI for identity validation and suspicious-activity detection. From the ITB experience, Amanin learned an important lesson: AI proctoring must be scalable. Technology that works well for only dozens of participants is not sufficient for university admissions conducted at national scale.
2. IPB University: Monitoring and Identity Verification Efficiency
The experience at IPB University provided a slightly different perspective. During the 2025 IPB Independent Selection, Proktor Amanin was used to help improve the integrity and efficiency of the examination process. One of the key requirements was to ensure that the person taking the examination matched the registered participant data. Proktor Amanin automatically verified participant identities by matching their faces with previously uploaded participant data.
According to IPB's testimonial, this approach was more efficient and carried a lower risk of error than manual validation. Interestingly, the IPB experience also demonstrated how AI can improve the monitoring ratio. In this implementation, a single proctor could effectively supervise approximately 65–75 participants. This shows that the value of AI lies not only in its ability to detect violations, but also in its ability to increase the productivity of human proctors.
3. Universitas Indonesia: Learning from the Complexity of Real-World Implementation
Universitas Indonesia provided another valuable experience. During SIMAK UI 2025 for undergraduate and vocational programs, Proktor Amanin was used to help reduce the risk of cheating, including the potential misuse of AI, while allowing participants from different regions to take the examination without having to travel physically to a specific location. However, the implementation also encountered technical challenges involving various parties—the institution, the system provider, and the participants themselves.
According to UI's testimonial, Proktor Amanin achieved approximately 80% of the defined objectives, including identity validation, suspicious-activity detection, and creating a more secure examination environment compared with an unsupervised examination. UI also stated its intention to use Proktor Amanin again for future postgraduate entrance examinations.
For Amanin, experiences like this are extremely valuable. Real-world implementations do not always work like demonstrations. Devices differ. Configurations differ. Network conditions differ. And user behavior cannot always be predicted. That is where a product is truly tested.
4. Telkom University: Operational Efficiency and Ease of Implementation
The experience at Telkom University highlighted another dimension of AI proctoring: operational ease. According to a testimonial from Telkom University, online proctoring helped improve the efficiency of examination monitoring, particularly for large-scale examinations, while reducing the need for physical resources such as large numbers of human proctors. The system also helped monitor suspicious behavior during examinations.
During the implementation with Telkom University, Amanin also received direct feedback from the IT team regarding ease of use, server stability, and the need to improve compatibility with Apple devices. For developers, feedback like this is extremely important. A good system is not simply a system with sophisticated AI. It must also be easy to use, stable, and compatible with users' real-world conditions.
5. FTMD ITB: AI Proctoring Is Also Useful for On-Site Examinations
The experience at FTMD ITB demonstrated that AI proctoring is not necessarily synonymous with examinations taken from home. Proktor Amanin was used for an examination involving 220 third-year students conducted in a physical examination room. The examination consisted of multiple-choice and essay questions, while the multiple-choice section used MS Forms, meaning participants still had to remain connected to the internet.
Although assistants were physically present to monitor the examination, monitoring the applications opened by participants had limitations when relying solely on human observation. Proktor Amanin therefore helped monitor:
- active applications;
- participant webcams;
- computer screen displays; and
- participant activity based on time.
This experience demonstrated that AI proctoring can serve as an additional layer of monitoring, even when examinations continue to be conducted physically.
6. Schoters by Ruangguru: Technology Must Be Supported by a Responsive Team
Schoters provided a perspective from the edtech environment. According to Schoters' testimonial, Proktor Amanin was considered beneficial for both participants and program organizers. One of the aspects that received particular appreciation was the responsiveness of the Proktor Amanin team, which made coordination with the Schoters operations team easier.
This experience demonstrates an important principle: In technology implementation, product and service cannot be separated. When an institution uses technology for a critical process such as an examination, it does not only need software. It needs a partner that can respond when problems arise.
7. Adeline Foundation: The Importance of Data for Post-Examination Review
The experience with Adeline Foundation provided a perspective on post-proctoring. According to its testimonial, Proktor Amanin was considered to offer comprehensive integration and features, including the ability to review records, camera footage, and screen activity during and after examinations. This data can be used again when suspected cheating occurs and a review process is required. This is important because not every decision has to be made while the examination is taking place.
Sometimes, an activity only becomes suspicious after the data has been examined more comprehensively. Therefore, a good proctoring system should not only generate real-time alerts. It should also provide evidence for the post-examination evaluation process.
From Field Experience Comes Mobile Pairing
One of Proktor Amanin's innovations also emerged from experience dealing with camera limitations. A laptop camera cannot see the entire room. Amanin therefore developed Mobile Pairing, which connects the participant's smartphone as an additional camera through a QR code. By combining the laptop and smartphone cameras, the monitoring perspective can be expanded to approach 360 degrees. This feature was developed, among other reasons, to reduce blind spots and detect the possible presence of other people outside the range of the primary camera.
This is an example of how implementation experience can generate new innovations. Problems encountered in the field become inputs for product development.
From Private Chat to an Integrated Help Desk
The same process occurred with the Help Desk feature. In an online examination, participants face more than just the risk of cheating. They also encounter technical issues. The camera may not work. The internet connection may be unstable. Participants may not understand the procedures. Or problems may occur with their devices.
Initially, Amanin provided a private chat between participants and proctors. In real-world implementations, this feature turned out to be widely used because participants needed assistance. Because many of the issues were technical in nature, the feature evolved into a Help Desk, allowing the proctor to invite IT Support into the same conversation.
As a result, examination security and participant support do not operate as two separate systems. Monitoring continues, while participant problems can be addressed without requiring them to leave the examination environment.
The Biggest Lesson: There Is No "One-Size-Fits-All" Solution
From these various implementations, Amanin has learned one important lesson: AI proctoring cannot be built using a one-size-fits-all approach.
ITB entrance examinations require the ability to handle thousands of participants. IPB requires efficient verification and monitoring. UI presents implementation challenges involving scale and complex conditions. Telkom University provides feedback on ease of use and compatibility. FTMD ITB demonstrates the need for application monitoring during examinations conducted in physical rooms. Schoters demonstrates the importance of responsiveness in supporting operations. Adeline Foundation highlights the importance of data and recordings for post-examination review.
All of these experiences have become part of the evolution of Proktor Amanin.
A Product That Evolves Together with Its Users
This is what differentiates simply building software from building a technology solution. Software can be created in a development environment. But a solution emerges when that software is used to solve real-world problems. The journey of Proktor Amanin demonstrates this pattern: User needs → technology development → implementation → feedback → improvement → next implementation.
The cycle continues. Therefore, Proktor Amanin's features—such as facial recognition, multiple-face detection, activity monitoring, AI proctoring, human proctoring, post-proctoring, Mobile Pairing, Help Desk, reporting, and integration with other systems—are not merely a list of features. Each feature has its own context and implementation experience behind it. Proktor Amanin is positioned as an AI-based online exam proctoring solution designed to safeguard the integrity and security of assessments.
AI, Humans, and Real-World Experience
Ultimately, Amanin's experience implementing AI proctoring leads to one simple conclusion. AI does not replace humans. AI strengthens human capabilities.
AI can monitor large numbers of participants. AI can record activities. AI can identify indications that require attention. AI can help verify identities. But humans are still needed to understand context, make decisions, and provide assistance when participants encounter problems.
Therefore, the success of AI proctoring is not measured solely by how intelligent its algorithms are. Success is also determined by how well the technology works together with humans and how effectively it can adapt to real-world conditions.
And that is the experience Amanin continues to build through various implementations with universities, companies, educational institutions, and other organizations.
From ITB to IPB. From UI to Telkom University. From FTMD to Schoters. From examinations involving hundreds of participants to examinations involving thousands. Every implementation provides a new lesson. Every lesson leads to an improvement.
And every improvement brings Amanin closer to one goal: To make online examinations not merely possible, but also secure, scalable, convenient, and trustworthy. Because ultimately, proctoring technology is not about how many participants can be monitored. Proctoring technology is about how much trust can be built in the results of an examination.