What we're thinking about next
A look at the directions Roma AI could reasonably grow in, based on where the product stands today.
This is a direction, not a promise. Nothing here is scheduled or guaranteed — items can be reprioritized, reshaped, or dropped entirely as the product evolves. Check /changelog for what has actually shipped.
Full dynamic pricing on every video model
Duration/quality/audio-aware pricing already covers Seedance and Wan — extending the same live-cost treatment to the rest of the video model lineup (Kling, Luma Ray 2, CogVideoX) instead of a flat per-model rate.
Roma Brain: a real conversational model
Chat currently exists as a surface in the product; the plan is to back it with an actual planning/reasoning model that can help scope a generation or a whole Workflow before you build it, not just hold a conversation.
Workflow template library
Starter Workflow graphs for common patterns (image-to-video, batch variations, multi-model comparisons) so a new project doesn’t always start from an empty canvas.
Mobile app
A dedicated iOS/Android app for generating and reviewing output on the go, sharing the same credit balance and history as the web app.
GPT prompt enhancement everywhere it isn’t yet
The Classic/GPT enhancement toggle already covers Image, Video, and Workflow model nodes — closing any remaining gaps as new surfaces are added.