AI in training centers: what to automate before September

Microsoft’s third annual AI in Education Report, published on June 24, 2026 from a PSB Insights survey of 3,345 people across six countries, has an uncomfortable finding for any academy or training center: 92% of students and education leaders and 88% of educators already use AI day to day, yet 77% of students and 53% of educators have received no formal training to do it. At Evicron, an AI and custom software studio based in Barcelona, we see the same gap in private training centers and academies: students and staff are already using AI on their own, almost always without the center having decided anything about it. With the new academic year around the corner, September is the moment to move from improvised AI to AI the center actually chooses and controls.

What Microsoft’s report reveals

Two figures from the report matter more than the rest for anyone running a training center:

• Adoption is no longer up for debate. With 92% usage among students and leadership, the useful question isn’t “should we adopt AI?” but “what criteria are we already using it under, and what criteria should we set going forward?”
• The training gap is the real risk. More than half of educators having received no AI training means day-to-day calls — which tool to recommend, what to grade, what counts as disallowed work — are being made with no shared criteria. Microsoft found that two out of three educators want monthly or quarterly training on the topic, not a one-off course.

That gap is exactly where a private training center can set itself apart from public education: with fewer layers of decision-making, it can set a clear AI-use policy and give its staff basic training in weeks, not academic terms.

Where AI fits into an academy’s daily work

You don’t need a massive project to get started. These are the three areas where applied AI pays off fastest in a training center:

Round-the-clock enrollment and inquiries

Most questions from a prospective student — schedules, prices, open spots, course requirements — are repetitive and arrive outside office hours. A conversational assistant connected to the real course calendar and available spots answers those questions instantly and, if the person wants to enroll, collects their details and completes the enrollment without anyone on staff having to step in at night or on weekends. As with any chatbot interacting with people, it should make clear from the first message that it’s an AI assistant: that’s good practice for transparency, and it builds more trust than hiding it.

Personalized content and learning paths

Generating level-appropriate variants of the same syllabus, creating review exercises, or adapting the pace of an online course to each student’s actual progress are tasks that today depend on how much time a teacher has spare. A well-scoped applied AI system — not a generic chatbot, but one trained on the center’s own syllabus and materials — can generate that supporting material under teacher supervision, freeing up hours for what a teacher does better than any model: guiding and correcting with judgment.

Tracking engagement and preventing dropout

In long or recurring-payment courses, spotting early that a student has stopped logging in or submitting work lets staff act before they drop out. Cross-referencing that usage signal with an automatic alert to the tutor — not to the student directly — is one of the applied AI projects with the clearest payoff in a center running multi-month courses.

How to get started before the new term

1. Pick one process. Enrollment, content, or engagement tracking: the common mistake is trying to automate all three at once. Start with whichever eats the most staff hours today.
2. Audit what staff and students already use. Before buying anything, ask what AI tools your team is already using on its own. It’s the fastest way to find your real starting point, in line with what Microsoft’s report shows.
3. Set a usage policy, even a short one. What can be asked of AI, what a person must always review, and what gets communicated to students. It doesn’t need to be a long document — one clear page already removes most of the risk.
4. Test with your own data, not a generic demo. An assistant trained on the center’s real syllabus and pricing gives useful results from day one; a generic demo only tells you whether to move forward.

Deliberate, not rushed

The right takeaway from Microsoft’s report isn’t “put AI everywhere right now” — it’s the opposite: AI is already inside your center, whoever is using it, and the open question is whether the center decides how it’s used or lets everyone improvise on their own. Setting that policy and automating one concrete process before September is a project measured in weeks, not a whole academic term.

At Evicron we work on applied AI for the education sector from a free discovery call through to production, with applied AI budgets starting at €6,000 and a working demo by week two of the project. If you want to figure out which process at your academy or training center is worth automating before the new term starts, get in touch: we reply within 24 hours.