Almost everyone in international education now says they are using AI. Far fewer can explain what it has changed in the way their organisation recruits, admits or supports students. The Koala’s recent coverage of the State of AI in International Education census captured this gap well, pointing out how much the sector talks about AI and how little it has measured.
One part of that census deserves more attention. It asks whether AI simply answers questions or is connected to other systems and able to take action. In my view, that distinction is the real measure of digital maturity, and it is the one the sector is least ready to discuss.
Using AI is the easy part
Anyone can open a chatbot. A recruiter can draft an email, an admissions officer can summarise a document and a marketing team can produce a week of captions. These are genuine time savers, and I would not discourage any of them. But they share a limitation, which is that the AI knows nothing about the student it is helping. It cannot see the application, the payment status or last week’s conversation unless someone pastes it in.
I think of this as standalone AI, a capable assistant sitting beside the work rather than inside it. Adoption surveys tend to reward it generously. Ask a hundred organisations whether they use AI and most will say yes. Ask how it touches a live application and the answers become much quieter.
Where the real work lives
Three systems sit behind almost every student journey. The CRM keeps the lead, manages recruitment, and processes admission. The SIS maintains the official student record that contains enrolments, fees, attendance, and results. The LMS contains the learning itself, including its content and assignments. In short, integration means these three systems share one student record, allowing them to share and understand what the other two know.
Picture three rooms in one house, with a student walking through all of them. In many institutions nobody has built a door between the rooms, so staff carry information across by hand. AI outside the house can only guess at what is happening inside. AI wired into it can act on what is actually there. In my experience, the technology is rarely the obstacle. The gaps between systems almost always are.
What integrated AI looks like
The disparity between the two becomes evident in daily activities. In the lead qualification process, standalone AI is capable of sending a courteous message, whereas integrated AI analyzes the request, verifies it and picks out the best candidates in a matter of minutes. In terms of application tracking, while standalone technology can create a common reminder, integrated technology can discern which documents are absent and request the students to furnish those missing pieces.
Routing and support follow the same logic. An application can reach the right admissions team the first time, without being forwarded around the office for a week. When attendance slips in the SIS and engagement drops in the LMS, a staff member can hear about it early enough to help, rather than after the student has withdrawn. None of this is exotic. It is plumbing, but plumbing is what turns a clever tool into a working one.
Where adoption gets overstated
Here is the uncomfortable part. Many institutions will report strong AI adoption, and some of it will be real. Some of it will be a handful of staff using personal subscriptions, which counts on a survey but changes very little about how the institution runs. Scattered systems aggravate the situation, as AI takes over any chaos that the system is connected to. A student might appear active in one system while being unnoticeable in another, and any decision made based on such an incomplete picture will be incorrect.
Students feel this long before leaders do. They repeat their story to three different teams and receive reminders about documents they sent weeks ago. Smart tools cannot fix poor connections.
People stay in charge
Integration is not an argument for handing everything to a machine. When AI links to functioning systems it can send messages, operate applications, and raise alarms, so institutions must have firm rules regarding responsibility. There will always be sensitive decisions like offers, rejections, welfare issues, or anything else affecting a student’s visa, which should be reviewed by a human being. Staff must understand how the AI operated and be able to change it if necessary.
The information regarding students is also sensitive and as such, privacy and access controls should be built into the design from the very beginning. Open-mindedness is a sign of maturity and being honest regarding the capabilities of artificial intelligence today.
What to measure instead
If usage is a weak measure, the sector needs better questions. How long does a new inquiry have to wait for a reply? How many requests have stagnated for lack of a document that was not followed up? How frequently does the same data get entered into several systems by the staff? What is the time frame for identifying the students at risk, and what comes next? Do students remain satisfied with the service provided? Monitoring this data in advance of an AI project and afterward will show if it can provide real benefits to the processes or it is just a nice decoration.
The divide ahead
The sector is heading for a split, though not between those who use AI and those who do not, since almost everyone will. The real divide will be between organisations whose AI is joined to the systems that run the business and those whose AI floats beside them. The first group will respond faster, spot problems sooner and spend less time on administration. The second will have impressive slides.
The census may give the sector its first honest picture, and I hope it asks the integration question loudly. The most useful thing any leader can ask right now is whether their AI is connected to anything at all.
Sanjay Laul is the founder of MSM Aventra, a unified CRM, SIS and LMS platform for institutions. He has spent over a decade working across integrated student lifecycle systems and AI-enabled workflows.
