TRUST CENTER
Responsible AI
Hisar is designed for auditable, sourced, human-controlled AI.
Sourced answers
Answers rely only on your own institutional sources and trace back to their source.
Refusal behavior
When it lacks a source, Hisar does not fabricate — it refuses. Appropriate refusal is a safeguard, not a defect.
No training on your data
Models are not trained on your data. Your data stays inside your boundary.
Human in the loop
For high-stakes outputs the human decides. Copilots like Nöbetçi accelerate the work; they do not make the decision.