Transparency
How we use AI
Every place AI touches your family's data, what it receives, and what a human still decides.
Last updated 28 August 2026
Our rule on AI
AI helps us mark faster and write more clearly. It does not decide anything about your child.
- A report is only ever produced from work your child has actually submitted. We never estimate, simulate or invent a score.
- Where evidence is missing, the report says so rather than filling the gap.
- Enquiry-stage documents are called an Initial Preparation Summary and contain no scores at all, because no work has been marked yet.
- A qualified human educator remains accountable for the educational judgement in every report.
These reports support your decision. They do not predict or guarantee admission to any school.
No automated decisions about your child
There is no automated decision-making with a legal or similarly significant effect. No model decides admission, selection, pricing or whether a child can use the service.
You can ask for human review of any AI-assisted analysis in a report, and for the reasoning behind a recommendation. Requests can be logged from the Data & privacy page in your dashboard.
Model training
We do not train our own models on your family's data, and prompts sent to providers are minimised to what the specific task requires — first names rather than full identities, no email addresses and no payment data.
Questions or concerns
If anything in a report looks wrong, tell us and we will re-check it against the underlying answers. Email sakuralearning1@gmail.com.
Feature-by-feature disclosure
Marking written answers in assessments, revision tests and exam papers
- Data used
- The question, your child's typed or handwritten answer, and the mark scheme.
- Provider
- Lovable AI Gateway (Google Gemini models)
- Human role
- Marks come from the mark scheme wherever the answer is objective. AI judgement is used for open answers and is reviewed before it appears in a report.
Report wording — Learning Profile, Blueprint, exam and interview reports
- Data used
- Your child's first name, year group, submitted answers and questionnaire responses for that report only.
- Provider
- Lovable AI Gateway (Google Gemini models)
- Human role
- Every figure in a report is calculated from real submitted answers. AI drafts the explanatory wording around those figures; a qualified human educator is responsible for the educational judgement.
Practice recommendations and next steps
- Data used
- Accuracy, topic history, targets and hint usage held in your child's record.
- Provider
- Lovable AI Gateway (Google Gemini models)
- Human role
- Suggestions are advisory. Parents and tutors can change or ignore any of them.
Interview practice voice and the website assistant voice
- Data used
- Only the text to be spoken aloud. No pupil identifiers are sent.
- Provider
- ElevenLabs
- Human role
- Voice synthesis only — it makes no assessment or judgement.
Website assistant (Forge) answering questions about products
- Data used
- The messages you type in the chat and which page you are on. Do not enter a child's personal details in the chat.
- Provider
- Lovable AI Gateway (Google Gemini models)
- Human role
- Sales and information support only. It cannot see pupil records.