Photo-led workflow
Inspiration image
- 01 Upload inspirationPhoto, screenshot, or saved reference
- 02 Understand the sceneDetect room elements and design items
Output: proposed item list + inferred room structure
Product workflowInterior design intelligence
Start with a photograph or a hand-drawn sketch. Confirm every important decision. Finish with a dimensionally grounded 3D room built from products the designer can confidently purchase.
01 / End-to-end flow
The entry experience changes, but both stories converge before product sourcing. Human confirmation is the control point that turns inference into an approved design input.
Path A begins with an inspiration photo and detects room elements. Path B begins with a hand sketch and style description, preserves the layout, and creates a styled concept. Both paths converge at designer confirmation, then proceed through product search, selection, dimensional verification, product-view preparation, 3D asset generation, room assembly, and final designer review.
Photo-led workflow
Output: proposed item list + inferred room structure
Sketch-led workflow
Output: styled concept + preserved spatial intent
Items · layout · room measurements · corrections
Find exact products first, then visually similar alternatives.
Exact + similarApply the designer’s country, currency, supplier, availability, and price settings.
Price-rankedThe designer accepts the exact match or chooses a suitable alternative.
Human-selectedExtract H × W × D, prioritizing supplier specifications over estimates.
Source-rankedFront, side, angled, rear, and top views where available.
Use generated views as internal reconstruction inputs when source views are missing.
Generate product geometry constrained to confirmed H × W × D.
Build the room shell from confirmed measurements.
Preserve the approved layout and real dimensions.
Move, rotate, add, remove, or replace products.
One trusted working model
Built from confirmed geometry, selected purchasable products, and retained evidence.
02 / Product stories
Each story preserves the designer’s authority while automating discovery, evidence gathering, and scene production.
As an interior designer, I want to upload a photograph of an interior I like so that the app can recreate its design using products I can confidently purchase.
As an interior designer, I want to upload a hand-drawn room sketch and describe the desired style so that the app can turn my concept into an accurate, editable, purchasable room design.
03 / Trust model
Every dimension and match keeps its provenance. Automatic inference can propose; only stronger evidence or designer confirmation can approve.
04 / Definition of done
Every included item was selected or accepted by the designer.
Every H × W × D value retains its source or confidence level.
The model has enough confirmed measurements to support its accuracy claim.
Supplier listings, availability, and product links remain accessible from the scene.