Managed Accounts

A managed account is a real social account you already own where AutoPersonas writes the copy and runs the schedule for you. There is no AI character, no image generation, no voice clone. You upload your own media or reuse imported posts, and the platform drafts captions in your real voice.

Managed vs Character profiles

Profiles come in two types and the choice is permanent at creation:

• Character: a fully synthetic AI influencer. Persistent visual identity, generated images, voice clone, the full pipeline.
• Managed: your real account. AutoPersonas only handles caption drafting, scheduling, dedup, and engagement. Your media stays yours.

Pick Managed when you have an existing audience and want help with the writing and posting cadence without changing what your feed looks like.

Getting started

1. Go to Influencers, New and choose Managed Influencer.
2. Connect a social account (X, Instagram, Threads, TikTok, Fanvue).
3. Run an Import to pull your last posts. Captions go into the voice corpus, media goes into the media library.
4. Compose a new post: pick media from the library, click Write with AI, review, and schedule.

Voice corpus

Every imported caption and every caption you publish through AutoPersonas is appended to your voice corpus, a per-profile collection of your own writing. The caption generator uses the most recent 20 entries as few-shot examples so new drafts sound like you.

• The corpus also powers duplicate detection. You will not be allowed to publish a normalized-text duplicate of a past caption.
• Near-duplicates (semantic similarity above 0.85) trigger a soft warning chip on the review item but do not block publish.
• You can see how many captions are in your corpus on the profile dashboard.

Media library

Imported photos and video covers from each platform land in your media library. Open the library page from the sidebar to browse, filter by source, and see how many times you have already reused a given asset.

From the post composer, click Pick from library to attach a previously published photo to a new post. The composer also prefills the original caption as a starting point if your draft is empty.

Caption generation for managed accounts

With an image attached in the composer, click Write with AIto generate a caption. The model receives:

• The image (multimodal grounding).
• Your last 20 voice corpus captions as few-shot examples.
• An optional topic or context line you provide.
• The most recent 10 captions, in an explicit avoid list.

If the first draft is a hash duplicate of an existing caption, the platform retries once with that draft added to the avoid list. If a usable, non-duplicate caption cannot be produced, you will see a clear error and can write the caption yourself.

Imports per platform

What is not in scope for managed accounts

• No image generation. You upload or reuse imported media.
• No voice clone or lip sync. Captions only.
• No reference packs, no wardrobe, no aesthetic grades.
• No automatic re-import. Use the manual Import latest posts button when you want to refresh.

Public API

Every managed flow is exposed via the public REST API. See Developer API for the full route list. Notable managed endpoints:

• POST /api/v1/social-import/{platform}, kicks off a historical post import.
• POST /api/v1/profiles/:id/generate-caption, returns a voice-grounded caption for a given media URL.
• GET /api/v1/profiles/:id/voice-stats, returns total captions in the corpus and the most recent samples.
• GET /api/v1/profiles/:id/media-library, lists imported media for a profile.