AI assistant · Getting started
How your assistant stays safe and honest
The assistant follows strict rules: it confirms before risky actions, only claims success when an action actually completed, and works within daily safety limits on emails, images, and destructive operations. Here is how those guardrails work.
It only claims what it actually did
The assistant is required to report outcomes truthfully. It only says something is done, sent, or changed if the action genuinely succeeded on your platform. If an action fails or cannot run, it tells you plainly instead of pretending.
It is also precise about money. For example, when it records a membership for a member you migrated from another system, it will spell out that no card was charged and no payment is scheduled — recording access is not the same as billing someone, and the assistant never blurs that line.
Confirmation before risky actions
For anything with real consequences — publishing or unpublishing a video, removing content, granting access, or sending email — the assistant shows you what it is about to do and waits for your explicit yes. Email always starts with a preview of exactly who would receive it and what it says; nothing is sent until you confirm.
Daily safety limits
Hard limits are enforced on the server, so they hold no matter what is asked: email is capped at 20 recipients per send and 100 emails per day; image generation defaults to 30 images per day and 500 per month; and destructive operations like deletions and cancellations have their own daily cap.
These limits exist to stop runaway loops and accidents, not to slow you down — daily counters reset at midnight UTC, and monthly image counters reset on the first of the month. If a limit is reached, the assistant tells you and explains when it resets.
Scoped to your platform only
The assistant can only see and change your own platform's data — your members, your content, your settings. It cannot reach other creators' platforms, and it cannot email addresses that are not members of your platform.
If it ever gets something wrong despite all this, use the Train button to send the conversation to the AutoCreator team.
Related articles
Meet your AI assistant
Your platform includes an AI assistant (Benny by default) that actually does things: it reports on your business, manages members, generates content and images, and emails members with your approval. Here is what it can do and where to find it.
Rename your assistant and set an avatar
The assistant is called Benny by default, but you can give it any name (up to 40 characters) and upload a custom avatar so it feels like part of your brand. Set it during signup or change it anytime in Settings.
Train your assistant when it gets something wrong
If the assistant gives a wrong or unhelpful answer, click the Train button to send the conversation to the AutoCreator team. Your submission is used to improve the assistant for everyone — no email goes to your members.