AI Disclosure

What the AI does

It interviews you during onboarding and records what it learns as memories.

It drafts post text in your voice, grounded in your own content and those memories, and shows you which memories it used.

It generates one image per post.

It proposes actions as cards you accept or dismiss. It does not act on its own.

What the AI cannot do

It cannot publish. There is no code path from the co-founder to a connected destination that does not pass through an approval and a consent record, and that is enforced by the system rather than by instruction.

It cannot reach another community's content or memories.

It cannot spend past the cap set for your community. When the cap is reached, generation stops with a clear message.

How AI-assisted content is labeled

Content the AI helped produce is marked AI-assisted inside the product, on the draft and on the record of what was approved.

When such a post is published, the caption carries a disclosure line stating that it was created with AI assistance.

Safety screening

Generated text and images are screened before they can be saved to the Studio. Screening reduces risk; it does not eliminate it, and it is not a substitute for you reading what you are about to approve.

Accuracy

Generated content can be wrong, out of date, or confidently incorrect. It's a draft. The approval step is the point at which a person takes responsibility for it, and that is deliberate.

Models we use

Drafting and image generation run on third-party foundation models, called through our own service so that prompts, temperature and structured output stay under our control. We do not send your content to a provider for training, and we do not permit providers to train on it.

Human oversight

Every published post has a named person who approved that exact version for those exact destinations, recorded against the version's digest. Change a character and the previous approval no longer applies.

Any channel can be switched off immediately by an operator.