Philosophy

A Room Isn't a Prompt. It's a Project.

The Idori team · August 31, 2026 · 4 min read

A room does not become finished because an AI generated one good image. Designing a space is a sequence of decisions, and the product should remember them.

The simplest possible AI design experience looks like this:

• Upload an image.
• Write a prompt.
• Generate a result.

That interaction is useful. We do not think it is enough.

Because your room is not a disposable input to a model. It is a project.

A generation has no memory

Imagine you upload your living room and ask for a warmer design. The result is good. Then you make another change. Then another. Soon you realize the first result did something better than the latest one.

In a basic generation interface, this becomes awkward. Which image did that? What prompt did you use? Can you return to it? Can you use that version as the starting point for another idea? Can you compare two results directly? Can you explore both instead of choosing one and losing the other?

Those are not edge cases. They are normal parts of making creative decisions. So we built Idori around a different model.

Every room becomes a project

When you upload a room to Idori, it becomes a project. From there, every applied change becomes a version in your Workspace.

That means the work has continuity. A result is not simply another image sitting in your camera roll. It has a relationship to what came before it.

Suppose you start with AI Ideas. Idori reads the room and presents five directions. You choose one. That creates a version. You decide the direction is right, but you want to tune it. You use Freeform Prompt. Another version. Then you use Spot Edit to change a specific detail. Another version.

The project now contains the history of your design thinking.

Version History changes how you experiment

When every generation feels disposable, people naturally become cautious. You hesitate to change something you already like. What if the next result is worse? What if you cannot get this version back?

Versioning changes that relationship. You can experiment without treating the current result as fragile. Try the change. If it works, keep going. If it does not, return to an earlier version.

This matters because experimentation is one of the ways people discover what they want. A design tool should encourage that, not make it stressful.

Compare instead of relying on memory

Human visual memory is not particularly good at evaluating small differences between two complex scenes. You may know that one version feels better without immediately understanding why. That is why Idori includes Compare.

You can compare versions directly and inspect the differences instead of trying to remember what changed three generations ago. Maybe the earlier version had better proportions. Maybe the newer one uses a material you prefer. Maybe one is clearly better once you see them together.

Comparison turns vague preference into a more deliberate decision.

One path is rarely enough

There is another problem with linear generation. Suppose version two is interesting. You continue from it and eventually reach version six. Then you realize version two had another possibility you never explored.

Should you abandon the current direction? Should you recreate the old result? Neither is ideal.

That is why Idori supports Branch. You can take a previous version and start a new project from it. One idea can become two directions. Those directions can evolve independently.

This reflects something we believe about design:

Creative work behaves more like a tree than a straight line.

You explore. You discover. You double back. You split directions. You abandon ideas. Sometimes you return to the idea you almost dismissed. The interface should support that reality.

The Workspace should preserve your thinking

We do not think Version History is merely a convenience feature. It is part of the product model.

The Workspace is where the design process stays visible. You can move through versions and understand how your room evolved. Each version is a decision point. Each can become a source for another refinement.

The goal is not to encourage infinite generation. The goal is to make exploration legible. By the time you reach a result you want to act on, you should understand how you got there.

Designing is not prompting

Prompting is an interaction. Designing is a process. That difference matters.

AI makes generating images incredibly easy. The next challenge is building products that help people use those generations meaningfully.

For Idori, that means treating the room as something persistent. A project. With versions. With history. With comparison. With branches. With room to change your mind.

Because the thing you are designing matters more than the prompt that happened to produce one version of it.