AI Data Handling (AskImpact)

AskImpact is Impact Suite’s AI assistant. It helps school staff draft and understand safety and wellness documentation. It is an assistive tool — a qualified staff member reviews and remains responsible for every output — and it does not make autonomous decisions about students.

The three questions districts ask, answered

1. Is student data used to train AI models? — No.
2. Where is student data processed? — Within Impact Suite’s AWS environment in the United States (us-west-2 / Oregon).
3. Who is the AI provider? — All AI runs on AWS services (AWS Bedrock with Anthropic Claude, plus AWS Transcribe, Polly, and Textract). No AI provider outside AWS receives student data.

Not used for model training

• AWS Bedrock does not use inputs or outputs to train the underlying models, and content is not shared with the model provider for training.
• Impact Suite does not fine-tune models on customer student data, and never sells, rents, or uses student data for advertising or profiling.

Where and how data is processed

• All AI processing (chat, retrieval, speech-to-text, text-to-speech, OCR) occurs within AWS in the United States — the same boundary that hosts the rest of the platform.
• Prompts are assembled from the minimum necessary data, drawn only from records the requesting user is already authorized to access.
• Retrieval (RAG) is scoped to a district’s own knowledge base; a district’s AI cannot retrieve another district’s data.
• Dictation audio is stored only transiently for the duration of a transcription job and deleted immediately afterward.

Human oversight

AI outputs are drafts for a qualified staff member to review. AskImpact does not autonomously act on a student’s record, placement, discipline, or services.

See the data flow for a visual of how data reaches these services.