Gym AI training: a step-by-step guide

Why does gym AI training matter? Because the inbox of a gym, a fitness center, or a pilates studio always fills with the same questions: "How much is monthly membership?", "Which days and times is the reformer pilates class?", "Is there a trial class?", "Can I work one-on-one with a trainer?", "Can I freeze it?". Group class, personal training, or an open-floor studio: it doesn’t matter; the same twenty questions repeat a hundred times a day. While the front desk tries to keep up with the phone and WhatsApp between classes, dozens of "price?" comments pile up under the post about the newly opened class slot. A prospect writes at 11 PM, "can I come to a trial class tomorrow," and waits until morning for an answer. Every membership question answered late goes to another gym.

inTusell takes over this load. But to do that, it first needs to learn your gym, your membership structure, and your class schedule. This article explains how a gym / fitness / pilates studio trains inTusell’s AI from scratch, step by step. This is the gym section of our sector-based "how to train your AI" series, and it’s the series’ core article.

Quick answer

Why is training important?

inTusell is an AI sales assistant and CRM that answers messages on WhatsApp, Instagram and web chat, books appointments and opens orders, and follows each customer through the sales pipeline; it is not a simple chatbot. It behaves like an experienced front desk and member greeter. A good greeter doesn’t know your gym on day one either. They have to learn which membership packages you offer, your class schedule, and which questions to leave to a trainer. Training makes the AI do two things right: give the right information (packages, class times, rules) and stop at the right boundary (not making up a price it isn’t sure of, leaving health and injury matters to the trainer and, when needed, to a specialist).

An untrained assistant either gives overly general answers or makes guesses about a subject it doesn’t know. In a gym, guessing is expensive. A wrong class time or an unfounded sentence like "this package suits you too" strings the prospect along, creates disappointment, and shakes trust in the gym. That’s why the most critical part of training is teaching the AI what it knows, and teaching it to hand over when it doesn’t know. A well-trained assistant stops when it doesn’t know.

Who is it for?

This guide is for sports businesses whose inbox is crowded with membership questions and class/trial requests, answering the same questions over and over:

• Fitness centers and classic gyms
• Pilates and reformer studios (high trial-class traffic)
• Yoga, crossfit, kickbox, and group-class studios
• Personal training (PT) and boutique training studios
• Teams that receive dozens of membership and class questions a day from Instagram and WhatsApp
• Businesses with multiple branches that update their class schedule often

You don’t need a technical team. All of the steps below are completed from sections in the panel, without writing code.

The structure of training: 5 core steps + 2 continuous steps

Splitting the training into two groups makes your job easier. The first 5 core steps get the assistant working; these steps are the one-time part of setup. The following 2 continuous steps (reviewing replies and turning fixes into rules, and opening channels) are the ongoing part you sharpen as you use the assistant. You’ll find them all in order below.

1. Upload the membership and class knowledge base

The foundation of training is the knowledge base (RAG). In the Knowledge Base section of the panel, you upload all of your gym’s textual information. The AI automatically searches these sources whenever it produces a response and uses only the information written here. In a gym, this directly means your membership packages and your class schedule.

Supported source types:

Every file you upload is automatically chunked and made searchable with pgvector. If you upload a "class schedule.xlsx" and a member then asks "what time is morning reformer pilates?", the AI finds the right row and states the time.

What must be uploaded in a gym: membership packages and their contents (duration, number of classes, included services), the weekly class schedule and instructor information, trial-class terms, freezing/cancellation/transfer rules, locker room and facility rules, working hours, and the most frequently asked questions. Important boundary: the knowledge base is a live source that must be kept current. If you changed a class time or a campaign ended, update the knowledge base: that is the foundation of a correct answer. Otherwise the AI assumes the old schedule still holds.

2. Set up trial-class and consultation appointments

The knowledge base explains "what you offer"; the appointment system lets the assistant give a prospect an actual time. In a gym, the most valuable return of this is the trial class and the intro meeting: the first step that moves a prospect to membership. In the panel you define three things:

Once these three are defined, the AI no longer gives times out of thin air; it proposes a trial class or meeting by looking at your working hours and booked times. The system prevents two appointments from landing on the same time. If you run fixed-time group classes such as reformer or yoga, you define them with capacity under Appointments › Settings › "Group Sessions" ("Tuesday 19:00 reformer, capacity 8", say); the AI offers those sessions with their remaining seats and never enrolls anyone in a full one. If Google Calendar is connected, one-to-one appointments are copied one-way to the team’s calendar; group-session bookings are not, they appear on the appointments screen. The prospect can reschedule or cancel the trial class themselves from the link in the reminder, which reduces last-minute empty spots.

Automatic reminders come into play here too: a reminder goes out one day before and two hours before the trial class, so fewer prospects fail to show up. A wasted trial slot is the most frequent complaint in gyms. Reminders run on paid plans. We’ve collected every detail of the appointment engine (durations, capacity, conflict prevention, reminders, and calendar sync) in a separate article: membership questions and trial/class booking automation. You can also see the membership and appointment capabilities on the solutions page and the gym solution page.

3. Set the persona and tone

You taught the AI what to say in the first two steps; now you’ll determine how it says it. There are two settings in the panel:

• Assistant name (ai_persona_name): The name it introduces itself with to the prospect. Most studios choose a name that fits their team: energetic but unembellished, reassuring.
• Tone (ai_tone): For a gym, the typical tone is warm, motivating, and fast; a greeter’s language that tries to understand the prospect’s goal (weight, flexibility, strength, post-rehab form) and moves them to a trial class without pressure. You can make the tone more corporate or more casual according to your own studio culture.

Multilingual note: whatever language the prospect writes in, the AI answers in that language. For messages arriving in English, Arabic, or Russian from foreigners living in your area, you do nothing extra; the assistant detects the language automatically and answers in it while preserving your tone, so the prospect is pointed to a trial class instead of slipping away.

4. Set the response rules and boundaries

The persona defines "who it is," the response rules define "how it behaves." In a gym this step does two jobs at once: it passes on your sales discipline and marks where the AI stops and hands over to the team. You add rules through the AI Manager Chat in plain language, just as you would brief a teammate.

Typical rules for gyms:

• "Before recommending a membership, learn the person’s goal, experience, and which days suit them; suggest the package and trial class that fit them."
• "Don’t close the exact price and campaign amount yourself; explain the suitable package and direct them to the trial class/meeting." (Adjustable according to your gym’s policy.)
• "Don’t state a class time, capacity, or campaign detail you aren’t sure of; say you’ll ask the team and hand over."
• "On matters like a health condition, injury, post-surgery period, or pregnancy, make no medical comment and don’t recommend a program; direct to the trainer team and, when needed, to a physician."

The most frequent mistake in a gym is the AI stating information it doesn’t have as if it were true: an out-of-date campaign, say, or a health-related promise like "this exercise is good for your lower back." In inTusell, two layers work against this. The main control is the system instruction, which constrains the AI from the start to use only the information in the knowledge base and to hand over when it isn’t sure. The second layer is an output filter: a pattern scan plus a second-model check that scans responses and flags misleading health or result promises. These two layers reduce the risk but are not a guarantee; the main safeguard is constraining the AI from the start, so that it doesn’t produce health advice or exaggerated promises, plus a current knowledge base.

If you want to compare two different approaches, you can use the A/B test feature: place a more informative closing next to one that calls more directly to a trial class, at 50/50 traffic, and measure which converts more trial classes.

5. Teach from past conversations

This is the step that takes training from "good" to "specific to your studio." In the panel you upload audio recordings of your past membership conversations (MP3, MP4, WAV, M4A) and mark each one as Won or Lost, for example, did the prospect come to a trial class and become a member, or were they lost.

The system uses these recordings in two ways:

So the AI learns how your best advisor uncovers the prospect’s goal, how they move the one who says "I’ll check the price and get back to you" to a trial class, and how they handle the "I don’t have time" objection. Personal data is protected: uploading a recording requires explicit consent, and phone, email, ID and card numbers are masked in the transcript before processing.

This step is not mandatory, but don’t skip it. The knowledge base teaches the AI "what" it knows; past conversations teach it "how you sell." When the two combine, the assistant truly resembles an experienced membership advisor.

A note: you keep slipping prospects warm with CRM and proactive follow-up. When a prospect asks for information and goes quiet, the AI writes up to 3 follow-up messages; a campaign for a new class period is something you start yourself, and it goes only to people who gave promotional consent. An opened spot is not offered to other prospects on its own.

6. Review and turn fixes into rules (ongoing step)

The first five steps get the AI working. After that comes the ongoing part that sharpens it over time. You check replies in AI Test and in Conversations. Say it explained a package’s contents incompletely or stated a class’s time wrong: you add the right fact to the knowledge base or write a rule in AI Rules. You can switch ready rule packs on or off in the Behavior Library, or give a plain-language instruction in the AI Manager Chat ("ask about their goal before giving a price," for example).

The Label Manager is a separate quality record: customer satisfaction replies, the outcome of a handed-off conversation, won/lost classifications and labels your team adds by hand land there; you approve, reject or edit them. These corrections do not change how the AI writes replies today; rules and the knowledge base change live behavior. If you want to see every reply before it goes out, use Full approval mode or the conversation’s Approval mode.

We recommend a short daily look in the first weeks. Review is the "live" part of training.

7. Open the channels (continuous step)

When training is ready, you put the assistant in front of prospects. inTusell gathers all channels in a single inbox: Instagram DM, Instagram comments, WhatsApp, Telegram, and Web Chat.

Channel-opening methods:

• Instagram: connects through Meta login; you pick your Facebook Page and Instagram account, with no keys or tokens to copy.
• WhatsApp: links with a QR code or with an 8-character linking code sent for your phone number; your number stays the same and no Meta Business approval is needed. The official line (Cloud API) opens after Meta approval.

In a gym, the first two channels to open are usually WhatsApp and Instagram DM, because that is where the questions from class posts and stories arrive. Instagram comment automation is live too. DM replies are already on once Instagram is connected; comment replies have their own switch. The "Instagram comments" switch on the Instagram card of the Messaging Channels page is on by default; if you like, you can turn it off, answer the comments under class posts by hand for a while, and switch it on when you are ready. If you connected Instagram before September 24, 2026, you need to reconnect it once for comments.

The account mode sets how much the AI does on its own (My business › Instructions › AI Settings › Advanced AI Settings): Intern (default) sends replies itself and asks you before a cancellation or refund; Senior handles those too; in Full approval every reply on WhatsApp, Instagram and Telegram waits for your approval. Most studios start with Intern; the AI handles membership and class traffic, and conversations involving a health condition or a special program go to the trainer team through Decisions. For the details of Instagram and WhatsApp automation: Instagram and WhatsApp automation for gyms.

How long does training take?

How long it takes depends mostly on whether your documents are ready. The steps of a working setup:

1. Uploading membership packages and the class schedule to the knowledge base (quick if your Excel schedule is ready)
2. Trial-class appointment types, group sessions and working hours
3. Persona, tone, and first response rules
4. Connecting channels

But there’s no moment when "training is done." In the first weeks, as you add the gaps you notice to the knowledge base and the AI rules, the assistant sharpens to your studio. Keeping the class schedule current and running tone A/B tests are ongoing improvements too. Setup is fast; mastery is continuous.

What it isn’t

It’s important to put inTusell in the right category, because a wrong expectation leads to a wrong setup.

• It is not a trainer or a health advisor. It doesn’t write exercise programs, doesn’t give injury/diet advice; it directs health and program matters to the trainer team and, when needed, to a specialist.
• It is not a membership management or turnstile system. It doesn’t keep member entry-exit records or manage membership terms and freezes; it’s a layer for communication, trial classes and sales follow-up.
• It is not a marketing tool that promises results. Phrases like "10 kilos in a month" or "guaranteed shape" are misleading, and the AI is constrained so that it does not produce them.
• It is not a bot that makes up information. It doesn’t state a price, capacity, or class time it isn’t sure of; it clearly says it doesn’t know and hands over to the team.

In short: not a bot that gives canned answers, but a member-greeting assistant that represents your studio to the extent you train it, but that doesn’t cross the health boundary and stops where it doesn’t know.

Frequently asked questions

How long does AI training take for a gym?

A working setup comes down to these steps: uploading membership packages and the class schedule to the knowledge base, trial-class and consultation appointment types, working hours, persona and tone, a few response rules. How long it takes depends mostly on how ready your documents are; the real refinement comes in the first weeks, as you add what you notice in AI Test and in real conversations to the knowledge base and the AI rules. Training is not one-off, it’s continuous.

Does the AI state the membership price or campaign amount?

It depends on your gym’s policy. The AI uses only the package and campaign information you upload to the knowledge base; it does not make up a price it isn’t sure of, it says it doesn’t know and hands over to the team. Many studios prefer to leave the exact amount to the consultation/trial class; you set this easily with a response rule.

Is technical knowledge needed for AI training?

No. All training is done from the panel; no code, API key, or developer is required. You upload membership packages and the class schedule as Excel/PDF, enter the trial-class appointment types and working hours, and write response rules in plain language. You connect Instagram through Meta login and WhatsApp with a QR code or a linking code sent for your phone number.

Does the AI create a trial-class or intro-meeting appointment on its own?

Yes. The AI creates a trial-class or intro-meeting appointment according to the class of interest and the member’s availability, taking your working hours and booked times into account; a double booking at the same time is prevented by the system. If you defined group classes as group sessions with capacity, it enrolls the person in a session that still has seats. In cases involving a health condition, injury, or a special program, it hands the conversation over to the trainer team.

Does the AI answer membership and class questions that arrive at night?

Yes. It answers nighttime questions on Instagram and WhatsApp like "how much is monthly, what time is morning pilates, is there a trial class" after hours too, and moves the interested person to a trial class. When the team arrives in the morning, prospects are warm and queued.

Are member and prospect data safe with respect to KVKK?

Yes. Data is stored encrypted and each business’s data is kept separate from the others; you can download your account data under Download my data (KVKK), and you manage the KVKK notice your customers receive, with your own privacy-notice link, under Legal notices. Uploading past call recordings requires explicit consent; phone, email, ID and card numbers are masked in the transcript. The AI never saves high-sensitivity fields such as a health restriction disclosure without your confirmation. If sensitive information like a health condition comes up in the conversation, the AI makes no medical comment; it leaves that to the specialist and the team.

Next step

You’ve trained your assistant; next is using it in daily operations. The next article in the series shows how the inTusell you trained runs on a real workday: a nighttime membership question, the trial-class schedule, reminders, no-show reduction, and handover to the team. Continue straight from there: how a gym uses inTusell.

This was the gym core article of our sector-based series. We built the same structure in adjacent sectors too; on the nutrition side you can check out dietitian AI training, for tourism tour agency AI training, and for health clinic AI training.

If you’d like to see it live before starting setup, use Get a demo for a short session where we open your panel together, or write to hi@intusell.com. For package and quota details see the pricing page, and for other guides browse the all posts list.