Last night, right after launch, we received a lot of user feedback and issue reports.

Most of it was around:

• Prompt understanding during image generation

• Aspect ratio inconsistencies

That feedback helped us quickly pinpoint the problems. We reviewed each issue, adjusted the logic, and as of today all reported issues are resolved.

Things are working more predictably now, but we’ll keep monitoring and iterating.

Sharing this because building in public + fast feedback loops genuinely help move the product forward.

On to the next improvements