AXHUB LECTURE · INTRO L3+

Booking and inquiry trades
— the first AI for academies, clinics, and real estate

Time about 30 minutesWhat you need one chatbot + 5 recent inquiries (chat, call notes, DMs)For academies, clinics, real estate, and other inquiry-heavy trades

This is a companion to L3 (review replies). An academy's enrollment inquiries, a clinic's booking inquiries, an agency's listing inquiries — the form differs but the structure is the same. It repeats, the answer is set, and if it falls behind the customer goes elsewhere. But these trades need one clearer line than reviews do — the line of "what AI must not answer."

1Why start with inquiry replies (3 min)

Inquiries across these three trades are a surprisingly identical repetition of questions. Schedule, cost, location, what to bring, availability.

The input is clear (the inquiry text), the answer is defined (your guidance), and mistakes can be fixed before sending — exactly the conditions of an easy-to-automate task.

Check: recall just three questions you got repeatedly last week.

2Make your 3-line rule (7 min)

Pick the closest example by trade and adapt it to yours. The key is the last line — the prohibition clause.

Practice 1 — examples by trade (pick and adjust)
[Academy] These are our academy's inquiry reply rules. (1) A warm greeting (2) The schedule and target grade for the subject asked about, specifically (3) One line inviting a consultation booking. Give only the tuition listed in the table, and for discount/refund rules don't give a firm answer — say "we'll explain in the consultation." Polite, three sentences.

[Clinic] These are our clinic's inquiry reply rules. (1) Greeting (2) Only hours, how to book, and what to bring (ID, etc.) (3) The booking link. Never answer questions about symptoms, diagnosis, or treatment — guide with "please come in and let us check during your visit."

[Real estate] These are our office's inquiry reply rules. (1) Greeting (2) Only the public info for the listing asked about (location, area, price range) (3) A visit-consultation suggestion. No firm answers on contract terms or final amounts, and don't invent listing details that don't exist.

What the three examples share: guidance is automatic, judgment and commitments are for a person. That distinction is the whole lecture.

Check: does your rule have at least one prohibition clause ("do not answer ___")?

3Five real inquiries — sort first, draft after (10 min)

Don't have it answer right away. Have it sort first.

Practice 2 — sort
[paste the step 2 rules]

Below are five recent real inquiries. Split each into "can be auto-answered" and "must be answered by a person," with a one-line reason for each.
[paste the 5 inquiries — with names and contacts removed]

Once sorted, pick only the "auto" ones and continue with "write a reply draft for each."

Check: is at least one of the five sorted as "a person's job" — if all are automatic, doubt your sorting criteria.

4The line you must not cross (5 min)

Every trade has a different red line. Pin it in a document and the line holds even when the part-timer changes.

Practice 3 — the always-a-person list
Academy: confirming refunds/discounts, grade counseling, a student's personal matters
Clinic: every question about symptoms, diagnosis, medication, or treatment; test-result inquiries
Real estate: committing to contract terms, price negotiation, explaining rights and encumbrances

Our shop's list: ____ / ____ / ____

Anything involving money, health, or law is a person's job — the criterion is the same across trades.

Check: did you post this list where the people you work with can see it?

5Save it as an FAQ document (5 min)

Don't throw away the drafts you made today. Gather them into a "frequently asked questions + our answers" document, and from next time it's paste-and-go.

This document is both for the AI and training material for new hires and part-timers — made once, used twice.

Practice 4 — save format
Document title: our ____ inquiry handling
1. The 3-line rule (+ prohibition clause) 2. The always-a-person list 3. Five FAQs with confirmed answers

If you want to learn more, free small-business support keeps opening — check the current round at Sobizz24 (sbiz24.kr).

Check: if the same inquiry comes tomorrow, can you reproduce today from this document alone?

When it doesn't work

It's a clinic and the AI tries to answer about symptomsMove the prohibition clause from the last line to the first — "Under no circumstances answer symptom or diagnosis questions; only give visit guidance." If it still leaks, change the routing so symptom inquiries never go to the AI and a person takes them directly.
The answer is stiff like a pamphletPaste in two replies you actually sent that got a good response and say "this voice is the standard." An example is faster than rules.
Inquiries are scattered across channelsChat, phone, DM, Naver — unifying channels is the next problem. Today, pick just the one channel with the most inquiries and start there. Once one works, the rest is copy-paste.

If review replies come first → L3. A shop owner's first AI · Full contents

Sources: criteria for automation fit (clear input, rule pattern) · government small-business support tracks (per official notices, verified 2026-07) · Stanford playbook (a person reviews only exceptions +71% vs +30%). Full sources are in the axhub.net case library. The practice examples are for following along; for medical, legal, and contract-related handling, always follow the relevant professional regulations.