With the global AI-in-education market projected to grow more than 36% in 2026, the challenge for school systems has stopped being “adopt” the technology and become integrating it as critical management infrastructure, one capable of acting on attendance, dropout risk, operational efficiency and, increasingly, student well-being, with governance, not just scattered tools.
A school that today collects attendance, behavior and performance data doesn’t have a technology problem. It has an infrastructure problem: what to do with that volume of signal before it becomes just one more forgotten spreadsheet. It’s this gap, between generating data and turning it into a decision, that sits at the center of the conversation about artificial intelligence in education in 2026.
According to a report published by E-Commerce Update, citing a survey by the consultancy Global Growth Insights, the global market for AI applied to education is expected to jump from US$18.6 billion in 2025 to US$25.4 billion in 2026, growth of more than 36% in a single year. It’s a consultancy projection relayed by a news outlet, not a closed or definitive figure, but it points to a change in nature: AI is moving away from being a one-off pedagogical support tool and being treated instead as a structural layer of school automation and management.
When people talk about smart AI monitoring in schools, the subject isn’t cameras or facial recognition. The scope is much broader: it covers attendance, dropout prevention, management’s operational efficiency and, in a still-early stage, student well-being, with physical security being just one of the possible layers, not the definition of the concept. This article covers that scope, with concrete cases already in operation for attendance, dropout risk and operational efficiency. And the question it answers isn’t “should we adopt AI in schools?”, that decision, for most school systems, has already been made. The question that matters is a different one: does the infrastructure, governance and trained staff exist to operate the data this AI is already generating?
What the 36% growth in the AI-in-education market is really saying
The number itself, US$18.6 billion to US$25.4 billion, according to E-Commerce Update’s reading of the Global Growth Insights study, is an investment statistic. On its own, it doesn’t tell much of a story. What makes it relevant is what it signals about market direction: increasingly, the AI platforms sold to school systems don’t arrive as an isolated tutoring app. They arrive integrated with management systems, automating triage, prioritizing alerts and organizing decision workflows that used to depend entirely on manual work.
That’s the shift that matters to whoever runs a school system: AI is increasingly being treated as infrastructure, something that underpins daily operations, not an extra resource that gets switched on and off. And all infrastructure, to work well, needs two things that no software license agreement delivers on its own: operational capacity to interpret what the data shows, and trained people to act on it. That’s exactly where most school systems are still behind the technology they’ve already bought.
From one-off tool to continuous infrastructure: what changes in practice
The difference between using AI software and operating a data infrastructure is the difference between putting out fires and preventing them. Standalone software answers a specific question: who was absent today, who has a low grade on this test. A data infrastructure crosses attendance, behavior, safety and performance in a continuous flow, and hands the decision-maker not a report, but a prioritization: who needs attention now, and why.
Two cases, one Brazilian and one American, show how this logic already works in practice, and both share a detail that isn’t incidental: the final decision stays human.
Attendance and dropout risk as a management signal, not a punishment tool
In the state of Rio Grande do Sul, the Department of Education, in partnership with the Evidence-Based Education Center, developed a risk-classification model that crosses attendance data with socioeconomic information to identify students at risk of dropping out. In tests run across the Porto Alegre school network, the model reached 91% accuracy in risk classification, a result measured against the tested dataset, not a guarantee of an individually correct call for every student, without using biometrics or facial recognition; the risk score is recalculated every quarter and feeds the state’s “Escola RS” system.
The report documenting this case dates from May 2025 and, while it falls outside the most recent window, the project continues expanding across the entire state network, with mentions at institutional events throughout 2025 and 2026 that indicate it’s still active.
What underpins the argument here isn’t just the model’s accuracy, it’s what deputy undersecretary Marcelo Brizolim described as technology’s real role: the AI prioritizes where the school counselor, whom he calls the “main actor,” should act first. The machine flags the signal. The person decides what to do with it.
In the United States, the Dunkirk City School District, in the state of New York, cut chronic absenteeism from 38% in February 2025 to 20% in February 2026 after deploying an AI solution that classifies absence risk across three severity levels, based on each student’s attendance history, and automates communication with families, with a response rate between 40% and 50%. According to superintendent Brian Swatland, the motivation for adopting the technology wasn’t a safety or compliance issue, it was the attendance team’s own workload, which spent more time manually compiling data than acting on it. The AI didn’t replace that team. It freed them up to do the work only a human does well: talking to a family, understanding a context, negotiating a student’s return.
Both cases point to the same principle: well-structured smart monitoring doesn’t take the decision out of the hands of the people who look after the student. It organizes the signal so that decision reaches them faster, and reaches the right person.
Student well-being doesn’t yet carry the same body of documented cases as attendance and dropout risk. Outside litigation contexts, there’s a shortage of public, verifiable examples of school systems using AI to monitor this dimension in a structured way, which makes it the next frontier of this same infrastructure logic, not an already-consolidated application.
The risk of growing without data governance: Brazil’s “silent pact”
Not every AI adoption in schools follows that script. An analysis published by the Escolas Conectadas platform, run by the Telefônica Vivo and “la Caixa” foundations, crossed data from a 2025 Itaú Foundation survey and arrived at a picture worth looking at closely: 84% of Brazilian students and 79% of Brazilian teachers have already used some AI tool in schoolwork. But only 32% of students received any formal guidance on responsible use of that technology.
That gap is what the analysis itself calls, without hedging, the central challenge of 2026, and it’s also the clearest picture of the risk this article wants to name. When adoption moves faster than the structure needed to operate it, AI stops being an advantage and becomes exposure: data with no defined architecture, decisions with no audit trail, use with no training for the people on the front line. Governance, in this context, isn’t an optional step to handle once the technology is already in use, it’s part of the infrastructure itself, as necessary as the AI model. It’s the difference between a school system that decides better with data and one that just accumulates technology without knowing what to do with it.
What separates a one-off initiative from real infrastructure
If the “silent pact” shows the risk, Piauí shows the other side of the equation. The state became the first territory in the Americas to include artificial intelligence as a mandatory subject in high school, with more than 120,000 students enrolled, 800 trained teachers and more than 540 schools offering the content.
The program, named “Piauí Inteligência Artificial,” won the UNESCO King Hamad Bin Isa Al-Khalifa Prize, competing against 86 initiatives from more than 50 countries, and runs with Google as a partner for platforms and services. The recognition was reported in October 2025 and continues to be cited as an active reference in more recent publications on the topic.
What makes this case relevant to the infrastructure discussion isn’t the curriculum itself, but what it reveals about method: planned scale, structured teacher training and a formalized technical partnership, not an isolated pilot at one school, nor a tool bought and distributed without preparation. It’s this kind of institutional architecture, not the amount of technology acquired, that separates a one-off initiative from infrastructure that sustains decisions on a continuous basis.
Connecting to educational infrastructure and continuous operation
The challenge these school systems face, organizing scattered attendance, behavior, safety and performance data into prioritized decisions, with the right people acting at the right moment, isn’t unique to education. Structurally, it’s the same challenge faced by any critical operation that needs to turn scattered signal into coordinated action. Operations centers, NOCs/SOCs and IOCs (Integrated Operations Centers), exist for exactly this: they receive alerts from multiple sources, apply a criticality criterion to each one, and dispatch the right specialist, without taking the final decision out of the hands of whoever has the context to make it.
This is precisely the territory Erione operates in: as a specialist in smart electronic security, structuring this kind of mechanism applied to AI monitoring in schools, from scattered attendance and behavior signal all the way to a prioritized decision. The overlap between these two worlds isn’t a coincidence: when the volume of data outpaces a human’s capacity to process it unassisted, what solves the problem is never more standalone technology, it’s infrastructure that receives the signal, prioritizes by criticality and dispatches whoever needs to act.
Conclusion: the future of smart AI monitoring in schools
The more than 36% growth in the AI-in-education market, projected by Global Growth Insights and reported by E-Commerce Update, isn’t the news itself. It confirms something the cases from Rio Grande do Sul, Dunkirk and Piauí, and the warning behind Brazil’s “silent pact,” had already shown separately: AI in schools has stopped being an experiment and become infrastructure, usually faster than the governance needed to operate it.
The question left for public managers, education departments and school operations directors isn’t whether AI will monitor attendance, dropout risk and operational efficiency, and, as the technology matures, well-being too. It’s already doing that, to a greater or lesser degree, across much of the school system, even if the well-being dimension is moving more slowly than the rest. The question is whether, behind that smart AI monitoring in schools, there’s infrastructure, governance and trained people ready to turn every signal into a responsible decision. Whoever answers that first won’t just have “more AI,” they’ll have a school system that decides better.
This is exactly the territory Erione operates in: smart electronic security applied to school management, connecting attendance, risk and performance into a prioritized decision, not an isolated camera, but infrastructure that turns data into action.
If your school system already collects attendance, safety and performance data, but still doesn’t have a structure to turn it into a continuous, prioritized decision, the conversation is worth having: well-planned management infrastructure is what separates adoption from results.