Here is exactly how AI works inside DetermiNext — what it sees, what it doesn't, and why every output stays in the teacher's hands.
The AI in DetermiNext has a narrow, specific role — and a long list of things it is explicitly not allowed to do.
Every step is server-side. No student PII ever reaches the AI model.
Student completes PINS assessment
Assessment responses are saved to the teacher's account on our backend. No AI call happens yet.
Teacher requests report generation
The teacher initiates report generation from their dashboard. The backend begins the AI request.
Backend strips all PII before the AI call
Names, IDs, and all personally identifiable fields are removed. Only anonymized PINS responses and career selections are sent.
Anonymized data sent to Google Gemini API
The stripped payload is sent server-side to the Gemini API. No browser or client code is involved.
Report returned and stored under teacher's account
The generated report is returned to our backend and stored. It is never transmitted back to Google.
Teacher downloads, reviews, and owns the final document
The teacher reviews the AI draft before using it in any formal planning context. Nothing is auto-applied.
AI output in DetermiNext is a starting draft — not a final document. No AI output is ever applied to a student record without teacher review.
Every report generated by AI is explicitly a starting point. Teachers are expected to review, edit, and approve all content before using it in formal IEP planning.
The teacher downloads the report and owns it entirely. DetermiNext does not submit, share, or forward AI output to any school system on the teacher's behalf.
AI models can produce inaccurate output. We recommend teachers treat AI-generated language as a structured starting point requiring professional review — not a verified clinical assessment.
If an AI generation fails or produces unsatisfactory output, teachers can retry generation at any time from their dashboard. No student data is lost on failure.
| Provider | Purpose | Data sent | Retained for training |
|---|---|---|---|
| Google Gemini API | Report generation | Anonymized assessment responses + career selections | No — prohibited by API policy |
Google's API usage policy prohibits using API inputs to train or improve their models. View Google AI terms →
If you're evaluating DetermiNext for district use and have specific questions about our AI implementation, we're happy to discuss the details.