How does a dietitian / nutrition consultant train the intusell AI? A step-by-step guide

A dietitian's or nutrition consultant's inbox is always full of the same questions: "How much is a session?", "Do you do online consultations?", "How many sessions does it take?", "Do you provide a diet plan?", "Should I come to the first consultation fasting, should I bring lab results?". Weight management, sports nutrition, pregnancy nutrition or clinical diet — it makes no difference; the same twenty questions repeat a hundred times a day. While you are in a client session, dozens of "price?" comments pile up under the recipe you just shared on Instagram, at 11 p.m. a prospect writes "how does online follow-up work" and waits until morning for an answer. Every consultation request you reply to late goes to another consultant.

intusell takes over this load. But to do so, it first has to learn your practice, your service structure and your working calendar. This article explains how a dietitian / nutrition consultant trains the intusell AI from scratch, step by step. This is the dietitian chapter of our sector-by-sector "how to train your AI" series, and it is the cornerstone article of the series.

Quick answer

Why does training matter?

intusell is not a ready-made chatbot; it is a fully autonomous AI assistant that behaves like an experienced client coordinator. A good coordinator does not start on day one without knowing your practice either: they need to know which services you offer, your appointment structure and which question they should leave to the dietitian. Training ensures the AI does two things right: give the correct information (services, packages, preparation instructions, working hours) and stop at the correct boundary (leave nutrition/medical counseling to you, promise no outcomes).

An untrained assistant either gives overly generic answers or guesses about a topic it does not know. In nutrition, guessing is risky: a wrong preparation instruction wastes the first consultation, and a boundary-crossing line like "with this diet you'll lose 8 kilos in a month" both misleads the client and shakes your credibility. Moreover, the AI does not have the client's lab results, chronic illness or medication; that is why it is not appropriate for it to make personalized nutrition recommendations. A well-trained assistant stops when it does not know and refers the client to you.

Who is it for?

This guide is for nutrition professionals whose inboxes are swamped with consultation requests and price/program questions, who answer the same questions over and over:

• Independent dietitians and nutrition consultants
• Weight management, clinical nutrition and metabolic counseling practices
• Those providing sports nutrition and performance counseling
• Those working in pregnancy, child and special-group nutrition
• Experts working primarily with online (remote) follow-up, taking clients from multiple cities
• Teams receiving dozens of consultation and price questions a day on Instagram and WhatsApp

You do not 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 make the assistant operational; these steps are the part of the setup done once. The 2 continuous steps that follow (label review and opening channels) are the ongoing part that you sharpen as you use the assistant. Below you will find all of them in order.

1. Upload the service and package knowledge base

The foundation of training is the knowledge base (RAG). In the panel's Knowledge Base section, you upload all the textual information about your practice. Every time the AI produces a response, it automatically searches these sources and uses only the information written here. For a dietitian, this means precisely your services, your packages and how you work.

Supported source types:

Every file you upload is automatically chunked and made searchable with pgvector. When you upload an "how online follow-up works" document, if a prospect asks "How is the program sent in a remote consultation?", the AI finds the right paragraph and responds.

What must be uploaded for a dietitian: service and package descriptions (first consultation, follow-up, online follow-up, package durations), how you work and the follow-up flow, first-consultation preparation (lab results, measurements, coming fasting, etc.), online consultation/payment/communication policy, working hours and the most frequently asked questions. Important boundary: the knowledge base is not for personalized diet or medical content. The AI uses the information here to answer "what do you offer, how do you work, how do I prepare" questions; it leaves questions like "I have insulin resistance, what should I eat?" to you.

2. Set up first-consultation and follow-up appointments

The knowledge base explains "what you offer"; the appointment system lets the assistant give a prospect a real time slot. For a dietitian, the most valuable use of this is the first consultation and the follow-up/check-in session — the first step that turns a prospect into a client. In the panel you define three things:

Once these three are defined, the AI no longer offers times out of thin air; it suggests a first consultation or follow-up based on your working hours and booked slots. Two appointments at the same time (double-booking) are prevented by the system — in a single-person practice this is especially important. When an appointment is created, if Google Calendar is connected, it is also copied one-way to your appointment calendar. The prospect can reschedule or cancel their consultation themselves via the /manage-appointment/{token} link sent to them; this reduces last-minute empty slots.

Automatic reminders also come into play here: a reminder sent one day before and two hours before the consultation lowers the no-show rate — a wasted session is dietitians' most common complaint. The same structure also covers follow-up sessions: a reminder goes out to a client who forgot their next follow-up consultation, so the program does not fall by the wayside. We collected every detail of the appointment and program tracking engine — durations, conflict prevention, reminders, no-show reduction and calendar sync — in a separate article: dietitian appointment and program tracking automation. You can also see the service and appointment capabilities on the solutions page and the dietitian solution page.

3. Set the persona and tone

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

• Assistant name (ai_persona_name): the name by which it introduces itself to the prospect. Most professionals choose a warm but professional, reassuring name that fits their practice.
• Tone (ai_tone): the typical tone for a dietitian is warm, calm and non-judgmental; a welcoming language that tries to understand the client's goal (weight, health, performance, managing a clinical picture) and moves them toward a first consultation without pressure. Because nutrition is a sensitive subject, sales pressure is kept low and emoji use is limited. You can adjust the tone to your own practice culture.

Multilingual note: the AI replies in whatever language the prospect writes in. For clients living abroad and looking for online follow-up, you do nothing extra when messages arrive in English, German or Arabic; the assistant detects the language automatically and replies in it while preserving your tone — a prospect who would otherwise slip away comes straight back to a first consultation.

4. Define the response rules and nutrition boundaries

The persona defines "who it is"; the response rules define "how it behaves." For a dietitian, this step does two jobs at once: conveying your working discipline and drawing the line where the AI should stop and hand off to you. With the AI Manager Chat, you add rules by writing in plain English, just as you would instruct a teammate.

Typical rules for dietitians:

• "Before suggesting a consultation, learn the person's goal and which days work for them; offer them the appropriate service and a first consultation."
• "Don't close on an exact session fee yourself; explain the service and direct them to a first consultation." (Can be changed according to your practice policy.)
• "Do not state a service, package or way-of-working detail you are unsure of; say you will ask the dietitian and hand off."
• "Do not give a personalized diet plan, do not give nutrition advice such as 'eat this, cut that'; do not comment on medical topics like illness, medication, pregnancy or lab interpretation, and refer to the dietitian."

The most common mistake in nutrition is the AI stating, as if it were correct, information it does not have (for example, a recommendation given without seeing the client's lab results, or a promise like "this diet is a perfect fit for your condition"). intusell runs two layers against this. The real control is this: by system instruction, the AI is constrained from the outset to use only the information in the knowledge base, to not provide personalized nutrition counseling, and to hand off when unsure. The second layer is a guardrail layer that scans responses and flags risky phrases (such as "guaranteed weight loss," "certain results," "guaranteed cure"); this layer runs in shadow mode by default, meaning it does not hard-block the response but detects and flags it. So the real safeguard is that the AI is constrained from the outset not to produce nutrition advice or exaggerated promises.

If you want to compare two different approaches, you can use the A/B testing feature: for example, you place side by side a more informative closing and one that more directly invites a first consultation, at 50/50 traffic, and measure which converts to more consultations.

5. Teach from past consultations

This is the step that takes training from "good" to "specific to your practice." In the panel you upload audio recordings (MP3, MP4, WAV, M4A) of your past client consultations and mark each as Won or Lost — for example, whether the prospect came to the first consultation and started a package, or was lost.

The system uses these recordings in two ways:

This way the AI learns how your best welcome understands the prospect's goal, how it moves someone saying "let me think about it" toward a first consultation, how it addresses the "can I trust online" hesitation. The KVKK side is protected: PII (personal data) masking and explicit consent are applied to uploaded recordings. When sensitive health-related expressions come up, the AI does not turn them into medical interpretation; they are used not as information but to learn your coordination style.

This step is not mandatory, but don't skip it. The knowledge base teaches the AI "what" it knows; past consultations teach it "how you coordinate." When the two come together, the assistant truly resembles an experienced client coordinator.

A note: intusell does not have a managed waiting list for keeping lost prospects warm. Instead, through CRM and proactive follow-up, you make a fresh offer to suitable WhatsApp prospects who previously showed interest and have communication consent when a new consultation slot or a suitable program period opens up. In other words, you build the "fill the freed-up slot with the next person" logic by reaching out again to permitted prospects.

6. Correct via label review (continuous step)

The first five steps make the AI operational. From there on is the continuous part that perfects it over time. Every AI response drops into a label review queue. Here you can do three things: approve, reject or correct.

The AI learns from these corrections. Say it described a package's scope incompletely or mis-described the online follow-up flow; you correct it, and in similar situations it now uses the right phrasing. Over time, patterns specific to your practice accumulate (the package names you use, your description of the first consultation, your style of directing to a first consultation).

We recommend spending 10–15 minutes a day on this queue for the first two weeks. During this period, the correction rate drops rapidly, because the AI learns the common mistakes. Review is the "live" part of training: the system gets smarter as it is used.

7. Open the channels (continuous step)

When training is ready, you put the assistant in front of prospects. intusell brings all channels together in a single inbox: Instagram DM, Instagram comments, WhatsApp, Facebook Messenger, Telegram, Web Chat and email.

Channel-opening methods:

• Meta channels (Instagram, Messenger): connected with one-click OAuth.
• WhatsApp: connected in about 1 minute by scanning a QR code from your phone — no Meta Business approval required. The official Cloud API option is also available.

For a dietitian, the first two channels usually opened are WhatsApp and Instagram DM, because the questions coming in under recipe and informational posts and stories flow there. Instagram comment automation, on the other hand, is rolled out gradually depending on Meta approval; the DM gate (instagram_dm_enabled) and the comment gate (instagram_comments_enabled) are separate. So you can first automate DMs and keep the comments under your posts manual, then bring the comment side into play once the comment gate is gradually opened.

There are working modes for handing off to a human: ai_only (the AI answers everything), human_only (everything goes to you), hybrid (the AI normally answers and escalates to you when needed). Most professionals start with hybrid; the AI handles the consultation and information traffic, while conversations requiring illness, medication or personalized nutrition go to the dietitian. A note: when a client sends a file, lab image, audio or video, that alone is not a reason to hand off to you; the AI handles the content appropriately and hands off only when medical interpretation or a personalized recommendation is required. For the details of Instagram and WhatsApp automation: Instagram and WhatsApp automation for dietitians.

How long does training take?

A working setup takes half a day:

1. Uploading services and packages to the knowledge base: 1–2 hours (shorter if your Excel list is ready)
2. First-consultation and follow-up appointment types and working hours: 20–30 minutes
3. Persona, tone and first response rules: 30 minutes
4. Connecting channels: 1–5 minutes per channel

But there is no moment when "training is done." Over the first two weeks, as you approve and correct responses in the label review queue, the assistant sharpens to your practice; keeping service and package information current and A/B tests are also ongoing improvements. Setup is fast; mastery is continuous.

What it isn't

Placing intusell in the right category matters, because a wrong expectation leads to a wrong setup.

• It is not a dietitian or nutrition expert. It does not write a personalized diet plan, does not interpret lab results, does not give nutrition/medical advice; it refers these topics to the dietitian.
• It is not a client-tracking software or nutrition program app. It does not produce diet plans, does not calculate macros; it is a layer for consultations, communication and client coordination.
• It is not a marketing tool that promises results. Phrases like "8 kilos in a month" or "guaranteed results" are both misleading and constrained so that the AI will not produce them.
• It is not a bot that invents information. It does not state a price, package or way-of-working detail it is unsure of; it clearly says when it doesn't know and hands off to the dietitian.

In short: not a bot that gives canned answers, but a client-welcome assistant that represents your practice to the extent you train it — yet does not cross the nutrition and medical boundary, leaving personalized recommendations to you.

Frequently asked questions

How long does AI training take for a dietitian?

A working setup is completed within half a day: uploading service and package information to the knowledge base, first-consultation and follow-up appointment types, working hours, persona and tone settings, a few response rules. The real refinement builds up over the first few weeks as you approve and correct responses in the label review queue. Training is not a one-off; it is continuous.

Does the AI give the client diet or nutrition advice?

No. The AI does not provide a personalized diet plan, does not give nutrition advice such as "eat this, cut that," and makes no medical interpretation; it refers these questions to the dietitian. The real protection is that the AI is constrained from the outset to use only the information in the knowledge base and to leave counseling to the expert; in addition, a guardrail layer scans for and flags misleading phrasing such as "guaranteed weight loss" or "certain results."

Is technical knowledge required for AI training?

No. All training is done from the panel; no code, API key or developer is needed. You upload services and packages as Excel/PDF, enter your first-consultation and follow-up appointment types and working hours, and write response rules in plain English. You connect channels with one-click OAuth or a QR code.

Does the AI create the first consultation or follow-up appointment on its own?

Yes. The AI creates a first-consultation or follow-up appointment taking your working hours and booked slots into account; double-booking the same time is prevented by the system. For matters involving illness, medication use, pregnancy or a special medical condition, it hands the conversation off to the dietitian.

Does the AI answer price and program questions that come in at night?

Yes. It instantly answers questions like "do you offer online consultations, how many sessions, how does follow-up work" that come in at night under Instagram posts and on WhatsApp, 24/7, and moves the interested person toward a first consultation. By the time you sit down at your desk in the morning, prospective clients are warm and queued.

Is client data safe with respect to KVKK?

Yes. intusell is end-to-end encrypted and KVKK compliant. Personal data you collect such as name, phone and goal is stored encrypted; when you upload past consultation recordings, PII masking and explicit consent are applied. All data is isolated on a per-tenant basis. If sensitive information such as a health condition comes up in a conversation, the AI makes no medical interpretation; it leaves that to the dietitian.

Next step

You have trained your assistant; next comes using it in daily operations. The next article in the series explains how to run the intusell you trained on a real workday: the night-time price question, the first-consultation calendar, follow-up reminders, no-show reduction and handing off to you. Continue straight from there: how a dietitian uses intusell.

This was the dietitian cornerstone article of our sector-by-sector series. We built the same structure in adjacent sectors too; on the health side you can check out clinic AI training, on the sports and fitness side gym AI training, for educational institutions education provider AI training and for tourism tour agency AI training.

If you would like to see it live before starting setup, with Get a demo let's open your panel together in a 20-minute session, or write to hello@intusell.com. For package and quota details you can look at the pricing page, and for other guides you can browse the all posts list.