Our
Story
Why We Built CoverGen
Every product starts with a real problem. Here's ours.
1
The Problem
Cover design was a daily bottleneck
Content creators were spending hours per cover — or
hundreds of yuan per design on outsourcing. The bigger the content library, the more the cost and time
multiplied.
2
The Research
Existing AI tools didn't solve it
General text-to-image tools produce random results.
Maintaining a consistent visual style — let alone consistent face identity — required complex LoRA
setups that few creators could manage.
3
The Insight
Reference-first generation changes everything
Instead of prompts, we use a reference cover. The AI
analyzes the style, then generates new covers that match it — with your face, your text, your aspect
ratio. No design knowledge needed.
4
Today
CoverGen — for creators and platforms alike
Built on FLUX.2, CoverGen serves individual creators
and large content platforms. From a single YouTube thumbnail to 10,000 covers/month via API — the same
quality at any scale.
The Pain We're Solving
These are the real problems our users had before switching to
CoverGen.
Hours spent per cover — time that should go into
content
Style inconsistency across videos — looks
unprofessional
AI tools generate faces that don't look like the
creator
Outsourcing costs ¥100–500 per cover at scale
CoverGen: Style from reference, face from photo, ready in
<15s for ¥0.03