How is Pre-Schematic AI Visioning different from Schematic Design?

Schematic Design produces plans, massing, and the first drawings of the building. Pre-Schematic AI Visioning produces no drawings at all. Its output is a shared direction on materials, light, and atmosphere, recorded as words in the brief, which Schematic Design then answers with an actual design.

Does Pre-Schematic AI Visioning replace the architect's design judgment?

No. The images are conversation tools. The architect chooses the vocabulary, reads the client's reactions, and decides what carries forward into the design. The judgment stays with the architect; the AI only makes a direction visible faster than a sketch or a stack of reference photos can.

Are the AI-generated images design deliverables?

No. They are discussion material, not representations of a buildable design. They do not address plan, structure, dimensions, building codes, or cost, and they should not be presented to the client as the final house.

What do I need to run a Pre-Schematic AI Visioning session?

An AI image generation tool, a clear picture of how the client wants to live, and precise vocabulary for materials, light, and viewpoint. Vague words return generic images, so the vocabulary matters more than the tool. Prompt Studio supplies that vocabulary as selectable chips and assembles it into a text prompt.

What is Pre-Schematic AI Visioning?

Q: Where can I see examples of images made for this stage?

The free WORKS gallery at tranexion.co.jp/works/ shows residential images organized by design nuance, region, and light phenomenon. It serves as public evidence of the image quality Prompt Studio is built to produce.

Pre-Schematic AI Visioning is the stage before Schematic Design in which the architect and the client use AI-generated images to align on materials, light, and atmosphere — so that design begins from a shared vision, not from assumptions.

The term describes a step that many residential architects and interior designers already practice informally, now that generative AI can produce a convincing image in seconds. Giving the stage a name makes it easier to plan, to explain to clients, and to keep separate from the design work that follows.

Where it sits in the design process

A typical residential project moves through these stages:

Brief and programming — how the client wants to live, the site, the budget.
Pre-Schematic AI Visioning — the direction of materials, light, and atmosphere is made visible and agreed.
Schematic Design — plans, massing, and the first drawings.
Design Development and Construction Documents.

Phase names differ by country. In North American practice, the stage falls between programming and Schematic Design. In RIBA Plan of Work terms, it belongs to Stage 1 (Preparation and Briefing), before Stage 2 (Concept Design). In every case, it ends where design begins.

The problem it solves

At the very start of a project, before any drawings exist, the image of the house has always been shared indirectly: other projects, magazine pages, a quick sketch, and a great deal of explanation. Both sides nod at something slightly different. The difference often surfaces only at the first presentation, after weeks of work.

Generative AI closes that gap. While the client is still describing how they want to live, the architect can put a clear image of that direction on the table, and refine it together a few minutes later, in the same meeting.

What it is, and what it is not

It is:

A shared direction on materials, light, and atmosphere, agreed before design starts.
Fast and iterative: an image can be revised during the conversation, not after it.
A record of vocabulary. The words that produced the agreed images become part of the brief.

It is not:

A design. The images say nothing about plan, structure, dimensions, building codes, or cost.
A rendering of the final house. Presenting a visioning image as the finished design creates expectations the project cannot keep.
A way to hand taste to a machine. It is a capability that belongs inside the architect’s own practice, alongside the sketchbook and the study model.

Three things to align

Materials

Name the material precisely. “Stone” returns a generic surface; “honed travertine” or “split-face limestone” returns a decision the client can react to. The same applies to timber species, metal finishes, and plaster textures.

Light

Light changes how every material reads. Agree on the quality of light, not only the time of day: diffused overcast light, low raking sun, the soft blue of early evening. A material that feels warm in one light can feel cold in another.

Atmosphere

Atmosphere is the overall character of the house: regional influence, mood, and the height of the eye. A Mediterranean courtyard and a Nordic interior can share the same materials and still feel entirely different. Viewpoint matters too; the same room seen from a standing eye level and from a seated one tells the client different things.

A simple session workflow

Listen first. Let the client describe how they want to live before showing any image.
Show two to four contrasting directions. Contrast helps the client say what they prefer, not just what they accept.
Watch the reactions. Ask what they respond to: the stone, the light, the openness, the planting.
Change one variable at a time. Keep the scene and swap only the material, or only the light, so the client can see what each decision does.
Record the agreed vocabulary. The words behind the images the client chose go into the brief and become the starting point for Schematic Design.

Why vocabulary decides the result

AI image tools follow what they are told. Vague words return vague images. What separates a generic picture from one that carries design judgment is vocabulary: the name of the stone, the finish of the metal, the quality of the light, the height of the eye.

That vocabulary is also what makes a session repeatable. The same words can be reused with the next client, adjusted, and refined over many projects.

Where TRANEXION Prompt Studio fits

TRANEXION Prompt Studio is a browser-based prompt builder for architects and interior designers, built for this stage. Instead of writing prompts from a blank page, you select materials, light, viewpoint, and regional character as thumbnail chips, and Prompt Studio assembles them into a structured English text prompt that you paste into your AI image generation tool.

Prompts and presets stay in your browser; nothing is uploaded to TRANEXION servers.
Subscription: $60/month or $600/year, with Paddle as Merchant of Record.
New subscribers can start with Your first session in Prompt Studio.

To see the image quality this approach can reach, browse the free WORKS gallery and its vertical edition, WORKS VERTICAL. Both are organized by design nuance, region, and light phenomenon, the same coordinate system Prompt Studio uses.

The residential images referenced in WORKS are AI-generated architectural visions. They do not depict any existing house, client, or project.