Most schools are not short of education technology. What they are short of is any link between the tools they already have.

Picture one student’s week. She learns a concept in class on Monday, practises it on one platform, sits an assessment on another, and gets extra help from a third app. Her teacher sees one slice of that. Her parents see another. Only she moves through all of it, and nobody is looking at the whole.

Now say she keeps getting algebra problems wrong. The practice platform’s AI spots the pattern and serves up more questions. The assessment system logs a poor test score. Her teacher notices she’s struggling in class, but has no idea what happened during her practice at home. Each system holds a useful piece. Unless someone sits down and joins them up, nobody finds out what’s actually going wrong.

That’s the part of the AI-in-EdTech rush that doesn’t get talked about much. A tool can get very good at understanding what happens inside its own walls and still know nothing about what happens outside them. For students, that can mean going around the same loop: answer questions, get a recommendation, do another activity, move on, with no one checking whether the real problem was ever fixed.

The tools aren’t useless. They’re just working on a learning process that doesn’t happen in compartments. Something a teacher explains on Tuesday shows up in Thursday’s homework. Homework can expose a misunderstanding that should change the next lesson. A later assessment shows whether the extra help worked. Cut those links and most of the context disappears.

AI needs that context to be worth much.

Where the explanation fits

Generating an explanation in seconds, or suggesting another exercise, is quickly becoming standard. The harder job is knowing where that moment fits in the student’s longer journey.

If a student has struggled with a concept three times, the useful question isn’t only “what should we ask next?” It’s whether she understood the original idea, whether the explanation landed, whether the trouble is limited to this topic, and whether her teacher should step in. Answering that means information has to travel between learning, assessment and teaching.

It changes how we think about data, too. Schools have collected academic data for years. Using it when it matters is where things tend to go wrong. A report that lands weeks after an assessment can tell a teacher what went wrong. Connected information might let her see the pattern while there’s still time to act.

The teacher shouldn’t be the glue

If a teacher has to log in to several platforms, compare reports that measure progress in different ways, and work out by hand who needs help, the technology is adding work instead of removing it. Good EdTech should make a teacher’s day clearer, not hand her another feed to keep an eye on.

None of this means everything has to live in one app. Schools will keep using different systems for different jobs, and that’s fine. The realistic aim is for those systems to cooperate well enough that what’s been learned about a student doesn’t vanish when she switches activity.

It matters most for personalised learning, which too often means nudging difficulty up or down, or changing the pace. Real personalisation depends on knowing what came before. Without that, an AI can tailor the next activity and still miss why the student was stuck in the first place.

A better question

So the question for EdTech shouldn’t be “where can we add AI?” The industry has spent plenty of time on what AI can add to education. A better one now is which existing parts of the learning experience AI could help connect.

That means looking past the chatbot, the recommendation engine and the automated marking. It means thinking about how information can follow a learner through instruction, practice, assessment and support, while teachers keep control of the decisions that count.

Connection brings responsibility. The more joined-up a system gets, the more carefully schools have to think about how student information is collected, interpreted and shared. More data isn’t automatically better data, and a prediction should never harden into a permanent judgement about a child. AI can make education more responsive, but only if it has enough context to understand the learner and enough restraint to leave room for the teacher.

It would also be a mistake to measure progress by counting how many platforms have bolted on an AI feature. A school doesn’t need every tool to get clever on its own. It needs the right information to reach the right person at the right moment in a student’s learning. That’s much harder to build, and it’s the part worth building.

Views expressed are personal

The author, Umakanta Rana, is Global President & CEO, TutorCloud AI