Does ibakepro train AI on my data?

Question

Accepted Answer

No. ibakepro does not train AI models, on your data or anyone else's, because it has no models of its own to train. Every AI feature in the product is a call out to a provider's API, and the answer comes straight back to you.
That means the honest version of "your data is not used for training" is a statement about our AI providers' enterprise API terms, not about something ibakepro polices in software. Below is what ibakepro can show you, and where the line is.
Where AI is used
AI sits behind a handful of features: analytics questions, social captions and campaign copy, cake design mockups and smart import column mapping.
Every one of them is a call to a provider's paid enterprise API. None is a consumer chatbot account, which matters, because consumer terms and API terms differ on exactly this question. Our AI providers are named at sub-processors.
Nothing runs unless you trigger it
No scheduled job, automation or background task in ibakepro calls an AI provider. AI runs when you press the button that runs it, and it is metered per use, so an unprompted call would show up as a charge you did not make.
What is stripped before it is sent
Two features remove identifying details before the call. Only these two:
The analytics assistant. Customer and company names are taken out of the results before they go to the model, along with contact details and addresses. The real names are put back into the answer as it reaches your screen, so you read a name the model never received.
Smart import. Contact details and long runs of digits are removed from your file before it is sent. Names and addresses are deliberately left in, because the model is being asked to recognise what kind of column it is looking at and it needs their shape to do that. Smart import records your consent before its first run.
Everywhere else, what you supply is what is sent. A social caption prompt, a question you type or a photo you upload goes to the provider as you wrote it or uploaded it. The redaction above is two specific features, not a blanket filter across the product.
What ibakepro keeps
One audit row per AI operation: a fingerprint of the prompt rather than the prompt itself, a short preview with emails and phone numbers stripped out of it, the model called, tokens in and out, the credit cost and whether it succeeded. Outputs are not stored at all. Each row expires after 90 days.
That record exists so support can see what you actually asked when you dispute a charge, and so there is a log of what went to a third party.
Where the claim rests
The parts above are things the software does, and you can hold ibakepro to them. Whether our AI providers train on API inputs is governed by their terms, and nothing ibakepro does can enforce it either way.
They publish those terms, and their enterprise API terms exclude training on customer inputs. If AI training is a real risk on your side, read them at the source rather than taking a vendor's summary, including this one. The processors ibakepro uses are listed at sub-processors, the contractual position is in the DPA and the handling commitments are in the privacy policy.
Where prompts are physically processed, and the one AI feature that leaves your region, is covered in where is my data stored.

Does ibakepro train AI on my data?

Where AI is used

Nothing runs unless you trigger it

What is stripped before it is sent

What ibakepro keeps

Where the claim rests

Run your bakery on ibakepro