Personal Chef Client Intake Form: Allergies, Diets and Menus With AI

A personal chef's intake form should capture five things: who is eating, every allergy recorded per person with the reaction and how strict to be, diets and dislikes, a kitchen equipment audit, and consent to store health information. AI can draft the form and a first menu, but you check every label yourself.

Most forms fail in one box: "Allergies: nuts". That answer doesn't say which nuts, what happened last time, whether traces matter, or whether anyone in the house keeps an adrenaline auto-injector. Because allergy answers are health information, the form also needs a storage habit and a rule about what never goes into an AI chat. AI is useful for two jobs here: drafting the form, and turning the answers into a first menu that you then check.

Follow me on Instagram@sagnikteaches

Household block: who eats, who decides and when you cook

Open with the facts that shape every later answer. A form that starts with allergy questions feels like an interrogation; one that starts with the household and the service keeps clients answering, and the allergy block then reads as the natural next step.

Connect on LinkedInSagnik Bhattacharya

If you build the form in Google Forms, use its "Go to section based on answer" setting on a multiple-choice or dropdown question. Ask "Does anyone in the household have a food allergy or medical diet?" as a dropdown with Yes and No, send "Yes" to the allergy section, and make the question required, so a skipped section never looks the same as a forgotten one.

Subscribe on YouTube@sagnikteaches

Allergy capture: one row per person and per allergen

Treat this block like a prescription pad rather than a survey: one row per person per allergen, never one free-text box for the household. Offer a tick list drawn from the 14 allergens named in some labelling rules (celery, cereals containing gluten, crustaceans, eggs, fish, lupin, milk, molluscs, mustard, peanuts, sesame, soya, sulphites and tree nuts). That list covers the groups on most shorter allergen lists. Add an "other" line for triggers that appear on no list, such as kiwi fruit or buckwheat.

The field that changes your behaviour in the kitchen most is strictness, and clients rarely volunteer it. Ask in plain words and let them pick one:

• Level 1: leave the ingredient out of the dish; traces and shared equipment are fine.
• Level 2: leave it out, avoid products labelled "may contain" it, and wash shared boards and pans first.
• Level 3: strict avoidance: separate or freshly cleaned equipment, no "may contain" products, and photos of every label sent to the household before you cook.

If two people in a household disagree, cook to the higher level and say that on the form. The levels are your own sorting aid, not a medical grade; a written plan from the client's doctor or allergy specialist outranks them. Precautionary "may contain" statements vary between manufacturers, so decide in advance how you treat them. The tutorial on precautionary allergen labelling builds a cross-contact risk log with AI, which is useful background for that decision.

Words get mixed up on the form, so ask which kind of reaction the client has. An allergy involves the immune system and can be sudden and severe. An intolerance usually causes discomfort and often depends on how much is eaten. Coeliac disease is an immune reaction to gluten that calls for strict avoidance. When a client isn't sure, record "suspected allergy" and cook to Level 2 until they bring clearer advice.

A single vague answer shows why the follow-up fields matter. This is an illustrative record, not a real client:

Two settings make the form behave. Make the allergy question required, with an explicit "none in this household" option. And end the block with a line such as "Tell me about any change before the next cook day; I will ask again every three months anyway", so the record never goes stale.

Diets, religious rules and dislikes: sort them before you plan

Clients use "diet" for five different things, and each one asks something different of you. Sort the answers into these groups so the menu draft doesn't treat a preference like a medical need.

The most useful question in this block isn't a diet label. Ask for five dinners the household already enjoys and three it has refused. Those eight answers give you, and an AI model, a far better brief than "likes healthy food". A fussy eight-year-old's list of refusals can save a whole batch-cook day.

The kitchen audit, item by item

An audit isn't about judging anyone's home. It tells you what to pack, how big a batch is realistic, and where cross-contact can happen. Clients fill in what they know on the form; you confirm the rest on a first visit or a ten-minute video walk-through.

For allergy households, add three questions to the audit: is the allergen kept in the house, where is it stored, and which shared items (toaster, colander, wooden spoons, non-stick pans) have touched it? In a coeliac household the toaster and the colander are the usual culprits. In a nut-allergy household it is the spoon that has been in the nut-butter jar.

A filled-in audit becomes a packing list. Take a kitchen where the audit finds an induction hob, one 30 cm frying pan, one small oven, no baking trays, two fridge shelves and no probe thermometer. Your pack list reads: two magnetic-base pans, three half-size trays, an oven thermometer, a probe thermometer and twenty labelled containers. The batch is capped at five dinners of four portions, because two shelves hold about that many containers. The form has done real work before you have chopped an onion.

From filled-in form to a first menu draft

Once the record is confirmed, strip it down to a profile with no names, no address and no medical letters, and give that to an AI assistant. The profile below is a worked example of that step for an illustrative household of four. Notice what the prompt forbids: it asks for a list of ingredients to check, never for a verdict that a dish is safe.

You are helping a personal chef draft one week of dinners for a household.
Use only the profile below. Do not assume anything that is not stated.

PROFILE (anonymised)
- Household A: 2 adults, 2 children (aged 7 and 10)
- Person 1 (child, 7): peanuts and all tree nuts. Level 3 strictness, including
  "may contain" products. Carries an adrenaline auto-injector.
- Person 2 (adult): coeliac disease. Strict gluten avoidance, including oats not
  labelled gluten-free and shared toasters.
- Dislikes: mushrooms, coriander, very spicy food.
- Five dinners they already enjoy: roast chicken, spaghetti bolognese, fish
  fingers and peas, cottage pie, vegetable curry with rice.
- Kitchen: induction hob, one small oven, no food processor, two fridge shelves.
- Service: one batch-cook day, five dinners, four portions each, eaten within
  three days. Grocery budget about $120 a week.

TASK
1. Draft five dinners that suit the whole household. One pan or one tray where possible.
2. Under each dinner, list every ingredient that could hide peanuts, tree nuts
   or gluten (sauces, stock, spice mixes, bread, pasta, oats, dressings,
   sausages) and write CHEF TO CHECK LABEL beside it.
3. Never describe a dish as safe or allergen-free. The chef checks labels.
4. Finish with a shopping list grouped by aisle and the equipment each dinner needs.

A realistic first answer, shortened here for space (illustrative output):

1. Lemon chicken and potato traybake with green beans
   Check label: stock cube (gluten), dried herb mix (nut warning), mustard
   Equipment: two large trays
2. Beef and bean chilli with rice
   Check label: chilli powder blend, stock cube, tinned beans in sauce
   Equipment: one large pan
3. Chicken stir-fry with rice noodles
   Check label: soy sauce (contains wheat; chef to find a gluten-free version),
   rice noodles (gluten-free label), sesame oil
   Equipment: wok or large pan
4. Baked salmon with sweet potato and broccoli
   Check label: marinade ingredients, any spice rub
   Equipment: one tray
5. Sausages, mash and peas
   Check label: gravy granules, margarine
   Equipment: two pans

Here is what I would change before anything reaches the client:

• The sausages themselves are missing from the check list. Many sausages contain wheat rusk, so for a coeliac household that is the first label to read, not the gravy.
• Dinner 1 needs two large trays and the kitchen has one small oven. The model ignored the equipment line. Either roast the potatoes on another day or halve the traybake and add a pan dish.
• Fish fingers, one of the household's favourites, never appeared. The draft reached for new dishes. Put two or three known favourites back in, because a menu the children refuse wastes a week.
• Some flags are noise. Mustard and sesame oil aren't allergens for this household, so cut them. Flags that don't apply bury the ones that do.
• The shopping list has no prices. The model can't know what your shops charge. Price it yourself, and if the total runs past the $120 budget, swap the salmon for a cheaper fish.

The rule that sits underneath all of this: AI can say where an allergen might hide, and you decide whether it is there. It cannot see the label in your hand, and recipes change between one batch of a product and the next. The same caution applies in restaurants, where whether a chatbot should answer allergen questions at all is the subject of the tutorial on AI chatbots and allergen questions. Read each ingredient list and any "may contain" line on the product you bought this week, not the one you bought last month, and write "label checked, date, initials" on the menu card.

When the menu is approved, the same profile can feed the next jobs. The tutorial on catering menu planning, with scaled recipes and shopping lists covers the arithmetic, and the shopping list the model produced above is a decent first draft to correct against your own price knowledge.

Keeping allergy answers safe: storage, consent and AI

Allergy records count as health information, and data-protection laws (including the GDPR, if you have clients in the EU) treat health information more strictly than ordinary contact details. I'm not giving legal advice here; if you hold records for many clients, ask a data-protection adviser to look at your set-up once. The practical habits below cover most of what a one-person business needs.

1. Collect only what you need. Allergens, strictness, written diet plans and the auto-injector location are enough. Don't ask for full medical histories or medication lists.
2. Add a consent line. For example: "I agree that [your business name] may store the allergy and diet information in this form to cook safely for my household, and will delete it within [period] after our service ends."
3. Lock the place where answers land. Use two-step sign-in on the account that receives responses, don't forward submissions to shared inboxes, and keep exports in one protected folder. Jotform's free Starter plan allows 5 forms and 100 submissions a month with paid plans from $39 a month ($34 on annual billing), and Google Forms is free; either works if the account behind it is secured.
4. Carry a one-line allergy card, not the whole form. One line per person is enough in the kitchen, and it keeps a lost phone or printout from exposing the full record.
5. Keep AI chats clean. Replace names with "Person 1", leave out addresses, doctors' letters and photos of medicine labels, and paste the profile rather than the form. Consumer ChatGPT and Claude plans have a model-training switch in privacy settings; business plans don't train on your content by default. Delete the chat when the menu is done.
6. Set a retention rule and keep to it. State on the form how long you keep records, and delete them when you stop working for the client.

The consent line belongs on the form; cancellation terms, liability wording and any allergen disclaimer belong in your contract, and a disclaimer never replaces the checks above. The tutorial on drafting a personal chef contract with ChatGPT lists the clauses to check.

If something goes astray, such as a form emailed to the wrong person, tell the client promptly and plainly. Trying to hide it costs more trust than the mistake did.

Build the form with AI, then test it on three awkward households

Drafting the form is a fair job for AI, and it takes about an hour including testing. Ask for the structure and wording, and keep the allergen logic as written above. A prompt that works:

Draft a client intake form for a self-employed personal chef who cooks in
clients' homes. Sections: household, allergies (one row per person per
allergen, with type, last reaction, strictness level 1-3, rescue medication and
written advice), diets and dislikes, kitchen audit, consent to store health
information. Use plain English, short questions and tick lists where possible.
Mark which questions must be required. Do not ask for full medical histories.

Expect these slips in the first draft, and fix each one:

• A single "allergies" text box for the whole household. Split it into one row per person.
• "List all medical conditions and medications". Remove it. It collects far more health data than you need.
• No strictness field, or a "severity" box with the choices mild, moderate and severe. Replace it with the three levels above.
• Questions that assume you control the client's kitchen. Reword them as "tell me about" prompts.

Then run three awkward test households through it before it goes live. Fill in each one yourself, or ask a friend to.

1. A family with one severely allergic child and a visiting grandparent. Does the form capture a guest who isn't in the weekly household? Can you record two different levels in one home?
2. A couple where one is vegan and the other coeliac. Does the menu brief keep both rules visible, and does the kitchen section ask about shared toasters and pans?
3. An older client whose daughter fills in the form. Is there a place to say who answered, and to record texture needs or professional guidance you aren't qualified to replace?

Whatever the form can't record cleanly, add. Send the first real submission straight back as a read-back message, so the client confirms what you have written down (illustrative wording):

Hi [first name], thanks for the form. Before I plan your menu, here is what I have
recorded.
Child 1, age 7: peanuts and cashew nuts, diagnosed, Level 3 (separate cleaned
equipment, no "may contain" products, label photos sent before I cook).
Adult 2: coeliac disease, strict gluten avoidance.
Auto-injector: red bag, hall shelf.
Please reply YES if this is right, or tell me what to change. I will not cook
until I have your reply. [your name]

Re-send the read-back every three months and after any change. That one message is the cheapest safety check in the whole process, and it doubles as dated evidence that you asked.

Questions chefs ask once the form is live

How do I handle guests at a dinner party?

Ask the host to send a short guest list with any allergies and diets at least 48 hours before the event, using the same one-row-per-person format as the main form. Treat anything unconfirmed as unknown, tell the host what you could not confirm, and cook the menu to the strictest level among the people attending.

What if a client won't share health details?

You can decline the booking or narrow the service. Many clients will say 'no allergies' and mean it, but a refusal to answer is different from a confirmed 'none'. Record the answer as 'declined to say', explain that you will then avoid the 14 common allergens where you can, and ask them to confirm that in writing.

What do I do if a client adds an allergy mid-week?

Stop cooking anything affected, update the allergy row, and send a fresh read-back message before the next cook day. Check which meals already in the fridge or freezer contain the new allergen, tell the household plainly, and remove them from the plan. Keep the dated change on file.

Does my insurance cover allergen incidents?

Don't assume so. Ask your insurer in writing whether your policy covers cooking in clients' homes and claims linked to allergic reactions, and whether it expects you to hold training certificates or written client confirmations. Keep the reply with your business records and revisit it whenever your services change.

Further reads

• Personal Trainer Intake Forms: Build One With AI, Protect Health Data — Another client form that collects health data, with the storage rules in detail.
• AI in Food Service: A Four-Week Menu Cycle With Costs and Allergens — Scale up from one household to a four-week menu cycle with costs.
• Recipe Costing Spreadsheet for Restaurants: Build It With ChatGPT — Cost each dish in the menu draft before you quote.
• Meal Prep Business Software: Weekly Menus, Orders and AI Labels — If your batch cooking grows, see what software handles labels.
• Event Planner Intake Form: Capture Every Detail Before You Quote — Borrow its approach to capturing details before you price a job.
• The Intake Questions That Make AI Cleaning Quotes Accurate — Another home-visit service where the right questions decide the quote.
• Horse Feed Chart Template: Build Yard Feed Sheets With AI — How one livery yard turned messy owner messages into a master feed sheet, change log and whiteboard copy, with prompts and sample rows.
• Wedding Catering Proposals: Tastings, Quotes and Follow-Ups With AI — A caterer's workflow for wedding proposals: tiered quotes you can check, tasting invites, guest dietary lists and follow-ups after the tasting, drafted with AI.
• AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.

Sources: published food-safety guides listing the 14 allergens named in some labelling rules; Google Forms and Jotform help pages (branching rules and plan limits); general data-protection guidance on health information.