Creator Studio v1
ActivePython/Flask prototype that proved AI influencers work. Eleanor Luxe—our synthetic influencer with hundreds of followers—validated the concept before we rebuilt everything for v2.
VisitOverview
Creator Studio v1—bare-bones infrastructure built from scratch. The painful lessons learned here, proven through Eleanor Luxe's success, directly informed v2's architecture.
How it works
Creator Studio v1 was a prototype that taught us everything. Flask framework with Celery workers and Redis queues, all hand-rolled infrastructure. Users updated Google Sheet templates—each row defining a day's post: activity, location, weather for outfit prompts. If a profile had a post scheduled that day, the system would generate and publish according to the posting schedule stored in PostgreSQL.
The database tracked everything: user profiles, scheduled posts, audit logs, generation settings, reference images for character consistency, published post history. ChatGPT Mini turned location and activity data into captions. Leonardo AI generated the images, using reference photos to maintain character consistency. Instagram OAuth built from first principles because we needed granular control.
Our proof of concept? Eleanor Luxe—an AI influencer with her own personality, posting schedule, and hundreds of real followers who don't know she's synthetic. The infrastructure creaked constantly. Flask wasn't built for this. Heroku's free tier wasn't meant for this load. But it worked, validating the concept and teaching us exactly what v2 needed to become.
Key features
- • Celery + Redis task scheduling for autonomous posting
- • Google Sheets integration for content calendars
- • ChatGPT Mini contextual caption generation
- • Leonardo AI image generation with upscaling
- • Multi-profile management for concurrent users
- • OAuth authentication built from scratch
- • PostgreSQL for profile settings and brand guidelines
- • Heroku deployment with worker pool balancing