Everything about what Bloom Kitchen AI is, what it will and will not do, how usage is measured, exactly what it costs, and how to stop it.
Last updated 2026-09-25
Bloom Kitchen AI is a kitchen assistant that lives inside your Bloomplate dashboard, right next to your recipes. Its job is narrow on purpose: it helps you use the recipes you already own, in the kitchen you actually have, on the day you are actually having.
That distinction matters more than it sounds. A general chatbot will happily invent a recipe for anything you ask, confidently, from whatever it absorbed from the open internet. Bloom Kitchen AI does not work that way. It answers from the Bloomplate recipes on your account. When the book has what you need, it names the recipe by its number and its title, so you can open it and cook. When the book does not have it, it says so, and offers the closest recipe you do own. If it ever suggests something improvised, it labels that idea plainly as not from the book, so you always know which is which.
Access to Bloom Kitchen AI comes with your book when you pay by card or PayPal. It is not one of the Bloom Weekly gifts, it is not an order bump, and it is not an add-on you have to buy. What you use is billed by usage, separately from the book, which is exactly why this page exists: you should know precisely what is measured and what it costs before you type a single word.
You never have to use it. Everything in every Bloomplate product is complete without it. The book stands alone, the Recipe Library stands alone, and every add-on stands alone. Bloom Kitchen AI is there for the moments when you have a question in front of the fridge and would rather ask than go searching.
The requests it handles best are small, concrete and immediate.
The pattern is simple. It is at its best when the answer already exists somewhere in what you own, and the work is finding it, adapting it to you, and handing it over in a form you can act on in the next few minutes.
You can attach a photo to a message. There are two things it is particularly good for.
"What can I make with this fridge?" Send a photo of your open fridge or your counter. Bloom Kitchen AI names the items it can see, asks you to confirm the list (a photo can hide a jar behind a jar, and it will not pretend otherwise), and then finds the recipes you own that use them. If you want, it will add those items to your kitchen list, which is what Bloom Daily Pick uses if you have switched it on.
"How can I make my bowl prettier?" Send a photo of something you made. Bloom Kitchen AI looks at it through the five BLOOM moves and gives you at most three specific fixes: a thicker base next time, a contrast color, a pattern instead of a pile, a finish at the last moment. It is kind about it. The goal is a better bowl tomorrow, not a grade.
A few plain facts about photos:
It will not give medical or nutritional advice. It will not tell you that a food treats, prevents or cures anything, and it will not describe a recipe as detoxing, cleansing, fat-burning or immunity-boosting. Bloomplate does not make those claims anywhere, and the assistant does not either.
It will not promise weight loss, and it will not turn a question about a recipe into a diet. If you ask whether something is good for losing weight, it will answer honestly and without claims: it can show you the recipe's estimated nutrition, labelled as an estimate, and it will suggest a qualified professional for anything personal.
It will not suggest an ingredient on your allergen list. That is not a polite request made to the model; recipes that contain your allergens are filtered out before the model ever sees them (more on that below).
It will not quote a product you have not bought. If the perfect answer to your question is a recipe in Little Bloom or the Protein Bloom Pack, and you do not own it, it will tell you which product it is in and leave it there. It will not paraphrase that recipe from memory. That is not a sales tactic; it is the same entitlement rule that governs your downloads, applied to the assistant.
And it has no ability to change your account. It cannot grant you a product, change a price, waive a charge, alter your eligibility or lift a limit. Those are decided by our system, in code, and Bloom Kitchen AI is not given a lever for any of them. If you ask it to, it will tell you plainly that it cannot, because saying anything else would be untrue.
Before every answer, the system selects the recipes that fit your question from the products on your account, and hands only those to the model. It is not browsing, it is not searching the web, and it is not drawing on a general pool of recipes.
Every Bloomplate recipe is stored as structured data: its title, category, time, servings, tools, ingredients, method, BLOOM moves, plating tip, tags, colors and allergens. Every recipe carries the id of the product it belongs to.
When you send a message, the system first filters in code. It keeps only recipes you own. It removes any recipe containing an allergen on your profile. It removes recipes that need a tool you told it you do not have. Only then does it match what is left against your words: titles, ingredients, tags, colors. The best handful, never more than eight, goes to the model.
Those recipes are handed to the model inside an explicit data fence, labelled as content rather than as instructions. That labelling is not decoration. It is what stops a recipe note, something you saved, or something you pasted from being able to give the assistant orders. If any of that text says "ignore your rules", the assistant treats it as text, not as a command.
It carries a small amount of durable context, if you give it: your first name, your household size, your diet (vegan, vegetarian, pescatarian or none), your allergens, foods you dislike, the kitchen tools you have, your favorite colors on a plate, and how confident you feel in the kitchen. That is the complete list. It is kept separately from your conversations, and you can view, edit and clear it from your dashboard.
Everything else is the recent conversation, which is trimmed as it grows. A conversation that grows without limit would be a bill that grows without limit, so only the last several turns travel with each message.
The practical consequence for your bill is that input is usually the larger side of a message, because the recipes it was given are counted too, and that caching does most of the work of keeping that side cheap.
Bloomplate is general home cooking. It is not medical or nutritional advice, and Bloom Kitchen AI is not a dietitian, a doctor or an allergist.
Allergies. Recipes containing your listed allergens are removed before the model sees anything, so it cannot suggest them. Every recipe also lists its allergens, coconut included. None of that replaces your own care: brands change recipes, and factories share equipment. If you mention a severe allergy, the assistant will remind you to check every label for cross-contact, and it means it.
Young children. Whenever you mention young children, the assistant applies the safety cuts from Little Bloom: grapes, cherry tomatoes and blueberries halved or quartered lengthwise, no whole nuts for children under five, no honey for babies under twelve months, and no hard raw chunks for toddlers. It will never suggest bee pollen to a child or to anyone with a pollen or bee-product allergy.
Food safety. When you are unsure whether something has gone off, the answer is to throw it out, and that is the answer you will get.
Nutrition. Where a recipe shows calories, protein or other figures, they are estimates, and the assistant will call them estimates. It will not build you a meal plan for a medical condition. For that, please speak to a registered dietitian or your doctor.
If you think someone is having an allergic reaction, contact your local emergency services immediately. Do not ask a chatbot.
Some recipes use a specialty ingredient, like pitaya powder or blue spirulina, and some work best with a particular tool. Your dashboard has a Shop page with the exact items we would use, and when you ask where to buy something, Bloom Kitchen AI can show you up to three of those items as cards in the chat.
It only does that when your question is about buying, ingredients or tools. It does not push products into answers that did not ask for them. The model never writes a link or a price: the cards are built by our system from a fixed list, and only for sellers on that list.
Some of those links are affiliate links. If you buy through one, we may earn a commission at no extra cost to you. As an Amazon Associate I earn from qualifying purchases. Every specialty ingredient also has a supermarket swap printed on its recipe card, so buying anything is never required. The full wording is on our affiliate disclosure.
This is the section most people come here for, so it is stated as plainly as we can manage.
Language models do not read characters or words. They read tokens, which are chunks of text roughly three quarters of a word long on average in English. A short question might be fifteen tokens. A paragraph of reply might be a hundred. The recipes handed to the model alongside your question are usually the largest part, and they are counted too. A photo is converted into tokens as well, and a larger or more detailed image produces more of them.
Every message therefore has two sides: input, which is everything sent to the model, and output, which is everything it writes back. They are priced differently because they cost differently.
Your price is derived from what the underlying AI provider charges us, multiplied by 20. That multiple covers the provider bill, the infrastructure around it, the recipe retrieval that keeps every answer grounded in your own book, the allergen filtering, the photo handling, and the safety rules described above. It is a fixed multiple, published here, and it does not change from one customer to another.
| What is counted | Per 1,000 tokens | Per 1,000,000 | When it applies |
|---|---|---|---|
| Input tokensEverything sent to the model: your question, a photo if you attach one, the recent conversation, and the recipes it was given. | $0.0400 | 40.00 | Everything sent to the model: your question, a photo if you attach one, the recent conversation, and the recipes it was given. |
| Output tokensEverything the model writes back to you. | $0.2000 | 200.00 | Everything the model writes back to you. |
| Cached inputInput served from cache, which is most of the fixed instructions on every message after the first. | $0.004000 | 4.00 | Input served from cache, which is most of the fixed instructions on every message after the first. |
| Cache writeThe first time a block of content is written into the cache. | $0.0500 | 50.00 | The first time a block of content is written into the cache. |
The rates above are the current customer-facing rates in US dollars, per 1,000 tokens of input and output. They are generated on this page from the live configuration, so what you are reading is what you would be charged right now, not a number typed into a document months ago.
After every answer, the AI provider reports exactly how many input tokens, output tokens, cached-input tokens and cache-write tokens that request consumed. We take those numbers, not an estimate. We do not count characters, we do not approximate from message length, and we never accept a token count sent by your browser, because a number that arrives from a browser is a number that can be edited.
Each category is priced at its own rate from the table above, added together, and written to your usage ledger as one row. When an answer needs more than one step (for example, finding recipes and then showing you shop cards), every step of that one answer is added together and recorded as a single row, exactly like a plain answer. The row records the model, the time, the four token counts, the provider cost, the multiplier applied and the resulting charge. The rate in force at the time is stored on the row itself, so if rates ever change, your history is not repriced. You can see every row in your dashboard under Billing.
Usage accumulates as an unsettled balance rather than being charged message by message. Nobody wants fourteen card transactions of a few cents each, and no card processor wants them either.
Settlement cannot fire twice for the same balance. Each settlement carries a key built from your account, the billing week and the exact amount, so a retry after a network timeout finds that key already used and does nothing.
It is easier to trust a number you can follow, so here is one, start to finish.
Suppose you open a new conversation and ask: "I have two ripe bananas and frozen mango. Something pretty for breakfast in ten minutes?"
Now ask a follow-up in the same conversation: "Make it for two." This time most of that input is identical to last time, so it is served from cache at the much lower cached rate, and only the new part is charged at the full input rate. The second message costs a fraction of the first.
At the rates in the table above, that first message comes to about $0.164, and the follow-up, with most of its input served from cache, to about $0.054. Your real numbers depend on the exact token counts the provider reports, which is why the row in your ledger, not this paragraph, is the authority.
This is the single biggest reason a conversation is cheaper per message than a string of one-off questions, and it is why we do not charge a flat fee per message: a flat fee would quietly overcharge exactly the people who use it most sensibly.
We could have published a flat price per message. We did not, for two reasons.
First, messages are not the same size. A one-line question about a swap and a long back-and-forth with a fridge photo cost very different amounts to answer, and a flat fee means the short question pays for the long one. Charging for what you actually used is the fairer arrangement, even though it is harder to explain.
Second, a published multiple can be checked in a way a flat fee cannot. The provider publishes its rates. We publish the multiple. You can therefore check our arithmetic, which is not true of a number we simply assert.
If you also have Bloom Weekly and Bloom Daily Pick running, they stay on separate lines with separate descriptions. A week might read, for example:
| Line | Amount |
|---|---|
| Bloom Weekly | $25.00 |
| Bloom Kitchen AI usage (an example amount) | $7.20 |
| Bloom Daily Pick, a normal week of 10 tasks | $16.50 |
They are never collapsed into one unexplained total. If you cannot tell what a charge was for by reading its description, that is a bug, and we want to hear about it.
A typical question and answer is a few hundred tokens of your own words plus the recipes it was given, and one or two hundred tokens of reply. In practice most questions land in the region of a few cents each once caching is doing its work. A photo adds to the input side of the message it is attached to.
Caching works in your favour. The fixed instructions and much of the recipe material are identical from one message to the next, so after the first message in a conversation most of that input is served from cache at the much lower cached rate. A longer conversation is therefore considerably cheaper per message than its first message suggests.
Someone who asks a handful of questions a week will usually sit under the $7 minimum for a while and see nothing charged at all for the first stretch. Someone who leans on it daily, with photos and long conversations, will see a settlement most weeks. Your actual number is in your dashboard, updated after every message, and you never have to work it out from this page.
There is one more protection: a daily ceiling of $20 of AI usage per account. Reaching it stops further billable use until the next day. It exists so that a runaway loop, a forgotten open tab or somebody else getting into your account cannot produce a bill you did not choose.
Bloom Kitchen AI is only available when there is a payment method on file that can legitimately be charged again. In practice that means a card, or a PayPal account you authorised for future payments at checkout.
If you paid with Apple Pay or Google Pay, there is a genuine technical reason it is not available: both wallets give merchants a single-use device number rather than a reusable card. There is nothing on file that can be charged a second time. That is a privacy feature of those wallets working exactly as designed, and it means anything billed by usage cannot be attached to them.
For those accounts, Bloom Kitchen AI is not hidden behind a message: it is genuinely off. The dashboard does not show it, our server refuses it, and no usage can be recorded. Everything you bought still works in full: your book, your Recipe Library and every add-on.
If you want access later, you can add a card deliberately from Billing in your dashboard. That is an explicit action you take, never something that happens because you clicked past a screen.
Before the first message that could ever be billed, you see a consent screen. It states that AI usage is billed separately, that usage varies with how much you use it, that the payment method you already have on file may be charged, what the current rates are, when settlement happens, where to see your usage and how to stop.
You tick a box. Nothing is pre-ticked. If you do not tick it, nothing happens and nothing is billed.
After that, we do not interrupt you again. A product that asks you to reconfirm every time you type is not being careful, it is being annoying, and it teaches you to click past warnings. You agreed once, knowingly, and the running balance is visible whenever you want it.
Under the chat box there is a permanent line: "AI Chatbot. Extra usage may apply. Learn more". The "Learn more" is the link that brought you here. It is on every screen where you can type to Bloom Kitchen AI, and it never disappears.
Your conversations are stored against your account so you can reopen them. They are readable by you, and they are not shown to anyone else. The system enforces that in two independent ways: conversations are keyed by your account, and every read checks again that the stored record belongs to the account asking for it.
Durable memory is deliberately small and deliberately named: your first name, household size, diet, allergens, dislikes, tools, favorite colors and kitchen confidence. Nothing else from a conversation is promoted into durable memory. A system that quietly collects everything you ever mentioned is a liability you never agreed to, so it does not do that.
Photos are used for one answer and never stored, as described in section 3.
You can view, edit and clear that profile from your dashboard, and you can delete any conversation. Deleting a conversation removes it.
Your food and household information is never sent to advertising platforms. Not your allergies, not your diet, not your kitchen list, not your household, not your photos, not anything you typed to Bloom Kitchen AI, not anything you set up in Bloom Daily Pick. Advertising events carry a purchase value, a currency and a hashed identifier, and that is the entire list. It is set out in more detail in our privacy notice.
Your messages, and any photo you attach, are processed by a third-party AI provider in order to produce an answer. They are sent for that purpose and no other.
Several limits run quietly in the background. They exist to stop a compromised account, a runaway script or an honest mistake from producing a bill nobody intended.
None of these affect anything you have bought. A paused assistant never removes access to a PDF, a recipe or your Bloom Weekly content.
Cards expire and banks decline things. If a settlement fails, the balance stays on your account as outstanding, and we try again in line with what the payment provider supports.
After repeated failures, billable AI use is paused. That is the honest option: the alternative is letting an unpaid balance grow on an account we already cannot charge, which serves nobody.
Your dashboard will say plainly what happened and what the outstanding amount is, and you can update your payment method there. When it goes through, AI use resumes.
Throughout all of it, everything you already bought stays yours. A failed AI settlement never removes access to a product, a download or your membership.
Stopping is simple, and there is no retention flow to click through.
Bloom Kitchen AI is built on a large language model, and large language models get things wrong. It is grounded in your own Bloomplate recipes specifically to reduce that, and it is instructed to say when something is outside what it was given rather than fill the gap. It will still occasionally be wrong. The recipe card is the authority whenever the two disagree, and your own eyes and nose are the authority on whether food is fresh.
Photo reading has limits too. It can miss an item at the back of a shelf or mistake one green thing for another, which is why it asks you to confirm what it saw before it plans around it.
Availability depends on a third-party provider. It can be slow or briefly unavailable. Nothing else in Bloomplate depends on it: your book, your Recipe Library, your shopping list and your downloads all work whether or not the assistant is up.
If a charge does not look right, or something on this page does not match what you see in your dashboard, write to us at support@bloomcookbook.com. Include the date and the amount and we will find it.
We would genuinely rather refund a charge you did not understand than keep it.