Aluna
An AI creation workspace that makes generation, comparison, and handoff feel more deliberate and controllable.
Lumen Labs set out to make AI creation feel deliberate rather than chaotic. I designed Aluna, a workspace where people can generate, compare, and hand off outputs with clear structure and control at every step. The product turns open-ended generation into a focused, repeatable creative process.
Year
Client
Work type
Prompt workflow
The prompt builder separates intent, references, and constraints so creators can steer the model without writing long, technical instructions. Structuring the inputs this way turned a blank text box into a guided, approachable starting point.
I designed the workflow to keep experimentation visible, making it easy to compare branches and return to a stronger direction whenever a path didn’t pan out. Nothing good ever gets lost, which encourages people to explore more freely.
Lowering the barrier
Many users arrived unsure of what the tool could do, so the interface offered gentle starting points and examples without dictating an outcome. That balance helped newcomers feel capable while leaving experts room to move fast.
By making each control map to a clear creative decision, the workflow taught people how to think about generation as they used it. The product became easier the more someone leaned into it.
Output review
Generated outputs were grouped by usefulness instead of by time, helping teams evaluate direction, fidelity, and next action at a glance. Sorting for relevance rather than recency matched how people actually make creative decisions.
Version notes and handoff controls made the product feel less like a black box and more like a genuine collaborative workspace. Everyone touching a project could see how it had evolved and why.
From draft to delivery
I designed the handoff step to carry context alongside the asset, so downstream teams understood the intent behind each choice. That continuity prevented the loss of nuance that usually happens between generation and production.
Clear provenance for every output also made review conversations faster and less subjective. Teams could debate direction against a shared record instead of relying on memory.
