context window — The maximum amount of text a model can consider at once — its working memory for the current conversation or task.
inference — Running a trained model to get answers — the phase where AI is actually used, as opposed to trained.
calibration — How well a model's confidence matches reality — a calibrated model saying "90% sure" is right about 90% of the time.
guardrails — The checks around a model that block bad inputs and outputs — filters, validators, and permission rules the model itself can't override.
Why it matters
Shieldstral takes moderation policies as plain-language input at inferenceRunning a trained model to get answers — the phase where AI is actually used, as opposed to trained.Full definition → time instead of a fixed harm taxonomy, so you can retarget guardrailsThe checks around a model that block bad inputs and outputs — filters, validators, and permission rules the model itself can't override.Full definition → per product without retraining. It runs on a single 16GB GPU with 3B parameters under Apache 2.0, covering text and images.
