Bodyguard’s role is to retrieve and analyze the content via our API, providing a verdict (HATEFUL, NEUTRAL, SUPPORTIVE) and recommended action (KEEP, REMOVE) based on that verdict.
The recommended action takes into account your own preferences, i.e. moderation needs/settings. For example, if an analyzed message comes up as HATEFUL but has LOW severity, if you have opted for letting low-severity content through, the recommended action would be KEEP rather than REMOVE.
We don’t remove the toxic content ourselves, we tell the platform what the recommended action is. Based on the API recommendation, the platform automatically keeps or removes the content, or can be configured to do other actions too - such as hide the content, store it somewhere etc.