Claude interprets references. Influencer OS structures the output.

Creators enter a character passport, scene requirements and references they are authorized to use. These become inputs to the prompt-building workflow; the application does not independently verify likeness rights.

Optional Claude features interpret reference images, suggest scene details, review prompt drafts and propose edits. The selected provider receives the content needed for each request directly from the browser, along with the user’s API key.

Influencer OS uses deterministic adapters to format the final engine-specific instructions. Identity and scene constraints are represented in the prompt, but the product cannot guarantee that an external model follows every constraint.

The creator uses separate image or video services to render the final media. Neither this website nor Claude produces those rendered assets in the current workflow.

Available in the released product: optional Anthropic Messages API (BYOK), visual-reference analysis, scene enhancement, prompt critique, assistant suggestions, deterministic adapters and licensed access.

Proposed, not shipped: repeatable evaluations for interpretation quality, prompt consistency, response latency and cost; explore an opt-in managed Claude experience without individual API-key setup.

Designed for transparently disclosed synthetic characters and authorized likenesses. Obtain permission for third-party photos and identities; respect platform disclosure requirements and provenance metadata. Do not use the product for deceptive impersonation.

Influencer OS is an independent project. Anthropic has not endorsed or partnered with it.