Never got an answer
The message came in at 9 p.m., in the middle of enrollment week. By the time someone read it, the family had already asked another school.
It answers questions about programs, class times, schedules and how enrollment works, books the trial class and follows up with the families who stopped halfway. It works on official WhatsApp, by text and by voice, at any hour — including the week when admissions season packs everything into a few days.
Pre-trained on your institution's website. Your team steps into the conversation whenever it wants.
In education, admissions are seasonal and the questions are always the same: curriculum, class options, schedule, course load, cost, how to sign up. The volume lands all at once, outside business hours, and the funnel leaks in places nobody can watch by hand.
The message came in at 9 p.m., in the middle of enrollment week. By the time someone read it, the family had already asked another school.
They asked for the curriculum, said they would think it over, and never came back. That isn't a lost cause — it's a second touch nobody had time to make.
The family went from "just asking" to "ready to enroll" in a single sentence — and that sentence got buried under dozens of other conversations.
A button flow can handle "browse our programs." It cannot handle "my son is nine, he's in morning classes — could he switch to the afternoon group?"
How many inquiries came in today? How many trial classes were booked? Who never heard back? The answer lives scattered across conversations.
Admissions season doesn't spread the volume out politely. Either the school answers while the interest is alive, or it answers later — to fewer people.
This isn't a chatbot running a flow. It's an operational agent trained to work.
It opens the conversation with a short voice note about your programs and takes it from there — in your institution's own vocabulary, using the content already published on your site.
Curriculum, class options, schedules, course load, cost and enrollment rules — grounded in the school's actual website, not in generic answers.
It picks up on intent, the student's profile, schedule availability and interest in a specific program — and records all of it in the conversation.
It schedules a time with the teacher or the coordinator, holds the seat and confirms on the day — without waiting for someone to remember to reply.
It circles back to families who never got an answer and to those who never finished enrolling, remembering what was already discussed — no starting over.
Sensitive topics — billing, academics, a special case — go to the right person, with the conversation history attached.
Text, voice and images, on the institution's official number through the Meta Cloud API. Sunday night is enrollment-decision time too.
The assistant recognizes the phone number of the principal, the coordinator or the head of admissions and switches roles. Facing the family, it serves. Facing the inside, it reports.
Until the integration with your student information system or school CRM is in place, every operational figure the assistant shows leadership carries an explicit simulation notice. It never claims to be connected to a system that hasn't been wired up yet.
Enrollment tracking is the WhatsApp conversation organized by stage: who just came in, who has a trial class booked, who still owes a document, and who is already enrolled. The front office opens the board and sees where each person stopped — and what has already been answered to them.
Four columns with a count of people in each stage: new inquiry, trial class, documents pending and enrolled. Each card carries the name, the class they asked about, the state of the conversation and the sentence the assistant already replied to that person — including who went quiet and when it will follow up. The table below repeats the same thing as a list, with the latest answer and the time of each one. The blue banner says plainly that the stages are simulated for as long as the academic system isn't connected.
During intake week, leadership always asks the same thing: who is stuck and why. Here the answer sits on one board instead of scattered across dozens of conversations somebody has to reread. And because the stage moves on what was actually said in the conversation, nobody has to remember to log the progress after hours.
The board isn't a form the front office fills in: the one writing on it is the same agent that answered on WhatsApp, with the memory of the whole conversation. That's why the card shows what was answered, and not just which box the person is in. When a subject goes beyond its reach — billing, academic, a special case — the handoff goes out with the history attached, and the stage records it.
Same number, same memory, different roles. Pick a side of the conversation.
The content depends on the programs, class options and rules each institution configures.
On the outside, the people who want to study. On the inside, the people who need visibility into admissions.
K-12 school, vocational program, language school or continuing education operation: the vocabulary and the sources it consults change, the structure doesn't.
The starting point is what your institution has already published. There is no flow to draw: you define the role, and the agent goes to work.
Programs, curricula, class options, schedules and published rules become the basis for what it answers. That's the pre-training.
Whatever isn't on the site — enrollment policy, exceptions, materials, the academic calendar — goes into the knowledge base it consults.
The school's number is activated through the Official Meta Cloud API and the assistant starts handling text, voice and images.
From the dashboard, a person joins the conversation and gets the context of everything already said. The agent goes quiet in that chat.
No. It handles the repetitive volume — curriculum, class times, course load, how enrollment works — and brings in the right person whenever the topic turns financial or academic. The conversation stays on the same WhatsApp thread, with the full history in front of whoever takes over.
It is pre-trained on the institution's website: the programs, curricula, class options and rules published there become the basis for what it answers. Anything not on the site is added to the knowledge base by your own team, with no developer involved.
Not out of the box. At first it works from the public content on your site and whatever the school adds. Reading student information systems, CRM and calendars happens through integration during setup.
Until a system is connected, every operational number it shows leadership carries an explicit notice that the figures are simulated.
The assistant introduces itself as the school's assistant and never pretends to be a person. When someone asks for a human, the conversation is handed off to your team and the agent goes quiet in that chat.
Yes. The same assistant serves K-12 schools, vocational programs, language schools and continuing education operations. The vocabulary and the sources it consults change; the structure does not.
Yes. Hal-AI agents serve customers in more than +40 languages — the same engine a major Brazilian airline uses for multilingual support over WhatsApp.
It does, and in that case the conversation starts from the other end: larger operations usually involve integration with in-house systems, service teams and conversation auditing. Talk to a specialist and we'll map out the scenario for your institution.
Interest arrives while your school is closed. The enrollment assistant answers on the spot, books the trial class and tells you what happened during the day.
School network, education group or an operation running in-house systems: talk to our sales team.