How a 250-Student School Runs on an AI Operations Team

The first organization that trusted Avelle with its operations couldn’t exactly refuse. My cofounder Anthony owns it.

We say this up front every time we talk about Académie Lavalloise, because it’s the honest frame — and because it’s the whole point. Before offering a managed AI operations team to anyone else, we wanted to run it where the stakes were ours: real payments, real families, real compliance obligations under Loi 25. If something broke, it broke on us.

The industry has a bad habit with case studies: vague clients, round numbers, no methodology. So here’s ours with the receipts — what the school looked like before, exactly what we handed over, and what changed.

The before: a capacity problem wearing a dozen disguises#

Académie Lavalloise is a non-subsidized private elementary school near Laval, about 250 students, run by a lean administrative team. If you read our first post, you know we sort every operational problem into four constraints: demand, conversion, cash flow, capacity. The school’s constraint was capacity — but on the ground, capacity problems never announce themselves that cleanly. They look like a dozen small fires:

Every response to a parent, every exam summons, every enrollment confirmation drafted by hand. Payments landing in the wrong records. Follow-ups that depended on someone remembering. Response times that stretched whenever the team was stretched. And over all of it, the traceability Loi 25 demands — a layer of rigor that’s brutal to maintain manually and consistently.

The standard answer is to hire. But for a school that refuses to push tuition upward — staying accessible to families is the mission — every administrative hire lands directly on operating costs. Growth ideas existed. The capacity to execute them didn’t.

What we handed over: twelve processes, not a tool#

Avelle took over twelve distinct operational processes, each running on a fixed schedule with strict validation rules. Not a chatbot bolted onto a website — the actual recurring layer of the back office:

The money coming in. Payment processing, financial monitoring, and late-payment tracking — matched, reconciled, flagged.

The money going out. Payable invoices, managed six days a week.

The families. Drafted responses to parent inquiries, the complete admissions process, seasonal communications, contract signature tracking.

The rhythm. A morning review every business day, a weekly action plan, a weekly review, and a continuously updated knowledge base so every draft speaks in the school’s voice.

Each process produces work ready to be validated — not work started from scratch.

The rule that makes it work#

One principle governs every sensitive process: Avelle drafts, a human sends. Every acceptance letter, every payment reminder, every response to a parent is prepared by the system and validated by a manager before it goes out. The machine absorbs the repetitive work; the human keeps judgment, context, and the final word.

That’s also why nobody at the school lost their job to this. The administrative team wasn’t replaced — it was reallocated to the work the school actually exists for: teaching quality, family relationships, development. Automation didn’t eliminate the important work. It freed the capacity to do it.

What changed — and how honestly we can measure it#

Before the deployment, those twelve processes represented an estimated 35 to 50 hours of manual work per week — the equivalent of a full-time administrative position. Today, the same work condenses into a few hours of supervision and validation.

Notice the word estimated. We built that range process by process — the time full manual execution would require at the school’s real volume — and we publish it as an estimate because that’s what it is. No stopwatch ran for a year. When a number is qualitative, the case study labels it qualitative. If a vendor gives you a precise ROI percentage on work like this, ask them for the methodology section.

The gains that don’t fit in a headline number turned out to matter most:

No communication ever goes out twice to the same recipient — the system tracks state persistently. Payments match the right records even when names vary. Late payments get flagged systematically instead of depending on human vigilance. Bulk mailings run in BCC so family addresses are never exposed to each other. And every automated action is logged, which turns Loi 25 traceability from a manual burden into a built-in property of the system.

Boring wins. The kind that compound every single week.

What a school’s back office proves for your business#

A school might feel far from your reality. The constraint isn’t. Professional offices, clinics, trades, service SMBs, manufacturers — anywhere recurring work piles onto a lean team, the same pattern holds: capacity gets absorbed by work that must be done but was never the mission.

What this deployment proved travels across sectors: an operation can handle growing volume without growing its structure. Errors of routine — duplicates, oversights, lost follow-ups — drop when state is tracked by a system instead of memory. The work happens every day, without absences or performance variation. And compliance stops being an afterthought when traceability is designed in.

Académie Lavalloise was the first deployment. It’s the same service we now run for SMBs across North America — proven on our own ground before it ever reached a client’s.

See the numbers for yourself#

The full case study breaks down the twelve processes, the methodology behind the estimate, and the compliance architecture — with every data point labeled as measured, estimated, or qualitative. If you’d rather talk about what your own recurring layer looks like, a thirty-minute conversation is enough for us to tell you honestly where we’d look first.

The question isn’t whether your repetitive work can be handed off. It’s how much capacity you get back once it is.

---

Matthew Lacopo leads client engagement and content at The Agensee, a managed AI operations firm based in Montréal serving SMBs across North America, which he cofounded with Anthony Lacopo.